Re: SUMMARY: FS#10849 - Sleep timer options: persistent duration and start on boot
snip all of the above sounds good. changes to Cancel Sleep Timer (hh:mm), showing the remaining time. all of the above sounds good, I do have a suggestion which I think could be useful for the rockbox voice users such as myself. I think its very good that the sleep timer menu will now show the remaining time, but the sleep timer menu takes several button presses to get to, so perhaps an option could be added so that you could set a voice announcement of the remaining time to the wps hotkey, or some combination of keys.
Re: Google Summer of Code 2011 ideas needed!
What about extending the aiff codec to play a wider variety of aiff files? I know rockbox supports playback of those files, but at the moment the files have to meet certain specific conditions in order to play. I have several files which for some reason rockbox doesn't like. I am not suggesting the idea just so I can play my files, but it would benefit everyone if the conditions for aiff playback weren't so specific.
Re: Google Summer of Code 2011 ideas needed!
On 06/03/2011 18:59, Mike Giacomelli wrote: What about extending the aiff codec to play a wider variety of aiff files? We already support something like 2 dozen AIFF formats. If you want to add another I don't mind, but I think this is more like a few hours work, not a GSOC project. aah, well seeing as I am not a coder, I don't really know what's involved, but it was just a thought. I just thought of it, as my files won't play under rockbox, and I am not sure why, I just presumed it might be because of the specifications in the manual which say how aiff files should be in order for rockbox to play them.
idea for formalising committal of new features.
Hi list, I have an idea for streamlining the committal of new features, and wonder what the list thinks. At the moment, it seems to me that committal of patches works on the basis that if someone with svn access is interested enough then a patch gets committed, but if it doesn't impact on people directly then nothing gets done or people seem to think well someone else will probably get around to committing that. So this is my idea, in edition to normal commiters being able to make changes to svn, Perhaps the release coordinator could have some kind of responsibility for reviewing new patches that introduce features, or if someone wants something putting in, the coordinator can be contacted directly about it. Basically I am suggesting that adding of new features needs to become more formalised rather than it just being down to anyone who has svn access and who feels like committing something. I think there needs to be a set procedure for requesting a patch be put in rather than just sending out a message to the mailing list, asking on the forums or asking on irc and hoping someone notices. I mean look at how many old open patches are on the tracker, I suspect if most of them were reviewed they are probably out of date, or no longer even relevant due to the length of time they have been open, as there are several on there that are several years old. I would suggest the reason a lot of them haven't been closed or committed is down to apathy and there being no formal procedures for managing the tracker. Perhaps the tracker should be reviewed once a year or something to decide what should be done about outstanding tasks. Anyway, these are just my thoughts. What do people think? Alex.
Re: idea for formalising committal of new features.
On 8/23/10, Dave Hooper d...@beermex.com wrote: I would say that it's about more than just committing it. If nobody with commit access cares enough about committing a patch, then who is going to maintain, bugfix, and document (and translate) that feature, and going forward ensure that the feature remains correctly documented and tested? I would suggest we don't need to formalise a procedure to get patches committed; but we might need to formalise something more on ownership after something has been committed. For patches that already include full documentation and no bugs, maybe all that's required indeed is agreement amongst developers that the patch gets committed and becomes an accepted part of rockbox trunk, but that might be less likely if it's a feature that developers themselves have no interest in. But also (speaking generally now) there are probably a number of patches that don't tick these boxes currently, and for those the trick might be to do all the work upfront and then try to convince devs that it is ready for inclusion. That's all very true, but at the moment, the tracker is a mess full of old patches, with nobody making a clear decision about what to do with them. The bug tracker is even worse than the patch tracker for this. The reason I suggest formalising inclusion of new features is because it seems to me that you can ask for things to be committed, and nine times out of ten, either there is no will to actually do anything even if people like the idea, or a few people say they don't like an idea and it gets rejected just on their say so.
Re: fs#10766
(I think it gets more complicated though if the skip mode also works when paused; if so, you probably don't want it to start playing when you hit-pause-to-cancel-binary-skip, so maybe a different method is needed... or just leave it with a hidden 5-sec timeout, really) I don't understand the above point can you please explain what you mean? I think consistency is important in rockbox. So, if pause was going to be used to reset the time, it should also pause the audio I think as well when pressed.
Re: fs#10766
I just mean the converse: if the music was already paused, you probably don't want to unpause it just to cancel the binary skipping. OK, well why couldn't you have a long press of pause would cancel? I don't think that would conflict with anything would it? so a short press would pause or resume the audio, and a long press would cancel. Alternatively, perhaps a long press could be used to undo the last skip? What do you think?
Re: fs#10766
I just mean the converse: if the music was already paused, you probably don't want to unpause it just to cancel the binary skipping. On 12 Aug 2010 18:19, alex wallis alexwallis...@googlemail.com wrote: (I think it gets more complicated though if the skip mode also works when paused; if so, you probably don't want it to start playing when you hit-pause-to-cancel-binary-skip, so maybe a different method is needed... or just leave it with a hidden 5-sec timeout, really) I don't understand the above point can you please explain what you mean? I think consistency is important in rockbox. So, if pause was going to be used to reset the time, it should also pause the audio I think as well when pressed. Hi, sorry for the double post. But I just had a thought, either a long press of play/pause could be used for canceling the timing, or alternatively as an undo skip command. But I also just thought of a third option. What about a quick double press on the button? Similar to a double click with a mouse button, that would be another option to explore for use with undo or canceling the mode.
Re: fs#10766
What happens if / when you make a mistake and jump the wrong direction? Well if you press the opposite direction to the one you intended, you are skipped in the other direction, half the distance of your previous jump, so of course you have to go back in the other direction to get back to where you want to be. Bare in mind though, that what is on the tracker isn't the latest version.
Re: fs#10766
What happens if / when you make a mistake and jump the wrong direction? Well if you press the opposite direction to the one you intended, you are skipped in the other direction, half the distance of your previous jump, so of course you have to go back in the other direction to get back to where you want to be. Bare in mind though, that what is on the tracker isn't the latest version.
Re: fs#10766
I think David Hall's point was about what happens if you press the wrong direction. E.g. let's say you're in the middle and you press right by mistake (you meant to press left). Subsequently, no amount of pressing left again will ever take you further left than the midpoint. You msut wait for the secret timeout and then start again (no 'undo') Oh yes, I see your point. Well, I guess maybe Sean is best qualified to answer that one, but how could you build an undo feature in? Given that a single press of left or right will do the binary skip in eather direction, and a rapid press of left or right twice quickly, will jump you to the next or previous track on the latest version.respectively. The only option I can see for an undo feature would be either a single or double press of up or down. What do you think?
Re: fs#10766
Well, one problem is that it's a modal interface without any indicator - I wouldn't personally mind having to wait 5 secs for the binary jump mode to finish but there's no visible (or audible?) way of knowing. A real undo would be limited to the availability of target keys (and their original meanings, which imo should not change for the sake of the binary jump mode being activated). So perhaps add Undo Last Skip to the context menu? That could be an interesting option. I do think Thomas is right, maybe a true undo would be making it a bit complicated, but I can also see your point of why an undo would also be good. Perhaps you could have some subtle click sound play just to indicate when search mode had timed out, as five seconds isn't a long time to wait at all really.
fs#10766
Hi list. I am writing to suggest committing fs#10766 this patch adds a new method of moving through audio files to the skip length option called binary skip. At first it does sound quite complex, but I have been using this patch for a while now, and routinely use it when I have a long audio book that I first start listening to on the computer, and then want to carry on listening to on my 140. Basically what the patch does is if your at the start of a file and you have binary skip enabled, is when you skip forward in the file, you are jumped to half way through the file. if you then go back or forward in either direction, you are then taken half the distance of your first jump, and each jump after that moves you in smaller and smaller amounts to enable you to home in on the area of the track you want to continue listening from. The idea is quite complicated to explain but it does make navigating around long audio files very easy. Also, Sean put in a few features that are not present on the latest patch in the tracker. One is that a rapid click of the keys that move you in the file will jump to the next track as they would normally do if binary skip wasn't turned on. Another useful feature is that voice announcements are made after every jump telling you where you are in the file. But I hope when Sean has time that this will be made configurable, as I feel if this is to be committed it should certainly be configurable weather to get the speech announcements or not. But I really like his concept, and it does add a very nice level of flexibility to long audio track navigation. I do also feel a jump to time feature would also be very useful,but I use fs#10766 almost every day when I am out and about. But jump to time would also be useful if I knew exactly where I wanted to get to.
question about setting up ccache with rockbox toolchain.
Hi list. I only recently realised that in order to take advantage of ccache when compiling rockbox I actually needed to configure it. I have a few questions that I hope someone might be able to help me with. Firstly, what sort of cash size will I need? At the moment, ccache is set to use the default cash size of 1 gb. Also, what is the number of files I should set the cash to hold? Next, because rockbox can be compiled for multiple targets, does each target that uses gcc need to have its own cash? I also have two versions of the rockbox code, in separate folders, one for working with patches and an unmodified version that has no changes, do these also need to have independent cashes, and if so how should these be set up? Lastly, what compilers in the toolchain do I need to set up ccache to use? At the moment, so far I have created one symlink just for gcc, but I assume I also need to create another for the arm-ebi compiler? how should I refer to it when creating my symlink? Many thanks for your help. Alex.
request for task closure.
Hi. I am wondering if someone could please close fs#10311 for me? As it doesn't look like its possible for a person who opened a task to close it. The reason I would like it closed is because since the patch was committed to change how list voicing works, this task is no longer relevant, as the implemented solution works a lot better than the old beep idea. Many thanks for your help, Alex. __ Information from ESET Smart Security, version of virus signature database 5331 (20100801) __ The message was checked by ESET Smart Security. http://www.eset.com
question about setting up build client
Hi list. I feel I am now ready to set my computer up to run as a build client for rockbox as compilation seems to work fine for me, using this virtual machine although I still haven't managed to shrink its virtual disc, I am still trying to work out how exactly to do that. But anyway, I have one question I hope someone might be able to answer for me. I believe I have set everything up correctly from a technical point of view. However on the wiki page it says Set -username to your IRC nickname, so we know who runs the client. I haven't ever used the rockbox irc channel and don't feel I am likely to do so, I prefer to use the mailing lists and I am on both the ordinary users list and the development list. However I am quite happy to provide a name, I was thinking of just my name as one word so alexwallis. So this is my question, is there some procedure I need to follow to register this as my official irc name? As I see all the big irc users have a page where there irc names are listed along with there real ones. As I say I doubt I will ever use irc, but obviously I guess someone needs to know what my irc name would be in case I should ever need to use it. Many thanks for your help. Alex. __ Information from ESET Smart Security, version of virus signature database 5287 (20100717) __ The message was checked by ESET Smart Security. http://www.eset.com
Re: assistance with logging output of make command to a file.
On Mon, May 24, 2010 at 10:08 PM, alex wallis alexwallis...@googlemail.com wrote: Hi list, Hi Alex , It logs the bit about generating dependencies and then just seems to stop logging. the command I have been using is make /rockbox/log.txt I have used the same output redirection method you used and it works for me. I'm just speculating, but could it be a path problem ? Did you try just typing : make log.txt ? Hi, sorry for the late response. the problem is the command isn't logging the actual errors, just the stuff that happens before they come up. however the error I am getting is something like cc1 the value is two large for the defined data type, that's very roughly what it is, it happens very early in the build process on some files in the tools folder. I wish I could have captured the output so you could see the exact error. But I have done some googling, and I actually now believe the problem is nothing to do with the rockbox code or how my path is set up. according to what I have read on google about very similar types of errors, the problem actually relates to how gcc works when it is working with folders that are being shared between windows and linux using cifs, or smbfs. According to what I have read, I need to put a few extra options in my /etc/fstab file, to enable some option called nounix and maybe another one called noserverino, I am having a little trouble adding the commands, but of course this is beyond the scope of this list, and I am sure I will get it figured out eventually with a bit of googling, it may just take me a few days. __ Information from ESET Smart Security, version of virus signature database 5147 (20100526) __ The message was checked by ESET Smart Security. http://www.eset.com
assistance with logging output of make command to a file.
Hi list, wondering if someone could please tell me how I log all output of the make command to a text file? the reason I want to know is because I have just set up a rockbox build environment running vinux which is a version of linux made accessible for the visually impaired out of the box, now based off ubuntu. It is being run as a virtual machine by the latest version of vmware work station. I am pretty sure everything is set up right, I used the rockboxdev.sh script in tools to install all the cross compilers, and then added them to my export path in my /etc/profile. When I generated the make file, I didn't get any errors or warnings. However, when I actually enter the make command I get a load of errors, and would like some assistance from the list diagnosing what the problem is. I have googled around, and I find plenty of sites saying how to log output from the terminal, however when I enter the command most of the sites seem to give, it logs some output to a text file but not the actual errors. It logs the bit about generating dependencies and then just seems to stop logging. the command I have been using is make /rockbox/log.txt Once I have all this set up and working I will then make a vm version and an installable iso available for anyone interested. __ Information from ESET Smart Security, version of virus signature database 5142 (20100524) __ The message was checked by ESET Smart Security. http://www.eset.com
Re: Google Summer of Code
snip. My GSoC project, as I mentioned, will mainly consist of porting the Flite text-to-speech engine for use in core Rockbox (and, hopefully, plugins). This will be quite the task and will involve, among other things, adding the infrastructure for loadable TTS engines and integration alongside the current voice system. That's about it, really. If you have any questions, feel free to ask (be it by email or in IRC). Thanks for listening/reading! Delyan Hi. I wish you good luck with your project, and would be very pleased if a tts does get integrated into rockbox. I am happy to help test out any code on target that kind of thing if it would be of use, I have an h140. __ Information from ESET Smart Security, version of virus signature database 5066 (20100427) __ The message was checked by ESET Smart Security. http://www.eset.com
Re: Proposal to update GCC for ARM
snip Are there any opinions? I don't use flac very much, so a slow down doesn't bother me if its not noticeable. however, one question, what about the irivers, do they free any ram or gain any binsize? __ Information from ESET Smart Security, version of virus signature database 4925 (20100308) __ The message was checked by ESET Smart Security. http://www.eset.com
problem using rockboxdev script to download cross compilers.
Hi list. I am wondering if someone could please help me with a problem I am having downloading the cross compilers using the rockboxdev script, and also answer a few questions regarding target cpus. firstly, are there only 3 types of cross compiler I need to download? as the wiki page on cross compilers talks about 3, but reading the window once the script is run is a little bit tricky with my screen reader, so that's why I am asking. I want to get all of the cross compilers, because eventually I want to get this machine running a build server it more than meets the recquirements. Now on to my problem I am having with getting one of the cross compilers. I was trying to get the sh-1 cross compiler, and I put in s for the target, but something must have gone wrong when the file was downloaded as I get a message about unexpected end of archive. So my question is could someone please tell me where I need to go in order to delete that corrupted file and start again? I realise the path is given where files are downloaded when the script is run, but it is not the easiest dialogue to read with a screen reader. So I would appreciate it if someone could tell me the path please and the file name to remove. Many thanks for your help. Alex. __ Information from ESET Smart Security, version of virus signature database 4650 (20091130) __ The message was checked by ESET Smart Security. http://www.eset.com
Re: setting up linux for compiling rockbox.
Hi, many thanks for your reply. The problem with the shared folder may this: The created folder is a virtual one and if you share this via vmware, it is still on the vm hard disk. Try to take the folder on your real OS and share it from there to the machine. Could you please tell me how I go about doing this? I understand about folder sharing in windows, but am not at all sure about how to do this under linux, or what I would need to do to get windows to share my rockbox folder with the virtual machine. Many thanks for your help. Alex. __ Information from ESET Smart Security, version of virus signature database 4650 (20091130) __ The message was checked by ESET Smart Security. http://www.eset.com
building accessible vmware images to work with rockbox.
Hi list. I hope this is the write list to post the following question, as it is to do with rockbox compilation. Through my connections at college, I have just encountered a person who has extensive experience of working with linux, and of making it accessible. I have been in touch with him and talked to him about rockbox as at the moment, for blind users the only option if we want to compile rockbox without help is to use cygwin or to switch to a propper linux environment. I would prefer to run a vmware image as then I can continue working in windows while stuff to do with rockbox is going on in the background. The guy I talked to seemed interested in assisting me with creating an accessible vmware image to compile rockbox, and also to turn my computer into a build server. He has suggested the best approach is to create an image from scratch, rather than trying to make the current vmware images accessible. So hear are my questions, if he assists me, what packages would he need to put into a distribution in order for me to compile rockbox? is it just the ones for cygwin, as in if he just puts in the packages that are listed for selection when you set up a cygwin environment would that be ok? if he follows the list given on the rockbox site I mean, are there particular versions he would need to put into a distribution. My next question is what packages would need to be put into a distribution in order for him to make me an accessible build client? as I have been keen to dedicate computing time to the project, but obviously couldn't til now because of only having cygwin, and not wanting to have to reboot my computer just to run the client. Last question, if he creates me accessible vmware images that can be used for compiling rockbox or running a build client, would there be any interest in hosting them so that other people could take advantage of them if they wanted to get rid of cygwin?
Re: building accessible vmware images to work with Rockbox.
Hussein, I also am seeing this as a possible good way to go for myself too. I work in vmware but I am unable to get a desent ubuntu or Debian based system going. Hi. I realise this is going slightly off topic, but just as a side note regarding your issue with not having a good vmware image to work with, have you tried vinux? the author of it is the guy who is interested in helping me. vinux is designed to be accessible out of the box, it includes braille support, speech and magnification. the author provides several versions, a live cd version, a USB pen drive version and a vmware image. I am hoping to meet with him next week to actually talk about how to put this accessible vmware image or images together for rockbox. if you are interested, the url for vinux is http://vinux.org.uk I am not sure what packages should be included to get a build client accessible, which is what interests me, and then of course I would like some kind of separate system to compile rockbox with for my own personal use. I am not sure if I was thinking of a script exactly to automate the process, I guess having some kind of gui would be quite nice as its more fun pressing buttons and selecting options than typing commands though I have a feeling it might be complex to put something together so perhaps command line is best. I guess also the commands should be quite similar to those used in cygwin.
question about r22419:
Hi list. I am wondering if someone might be able to answer the following question for me about r22419: which adds a feature to the rockbox info screen displaying the amount of space used by the skin buffer. the notes for the revision say show the skin buffer size as well as the used amount in the rockbox info screen I am running rockbox on an h140, I compiled a fresh build today from the latest source and also generated a new voice. In the rockbox info screen I am seeing information telling me the skin buffer size is 6.44 kilobytes. However I am not seeing any information saying how much of that is used. Is that normal for the h140? or is this information simply not being voiced? Sorry for the question, but I am just slightly confused as it sounds like I should be seeing two pieces of information in the rockbox info screen about the skin buffer not just one.
Re: question about r22419:
pray tell... how? doing it the same way as the disk info adds a useless string, and really if you are relying on voice then there is a damn good chance you wont be loading big themes so it wont matter Hi. Following that argument, why bother voicing any of the visual information under rockbox info? such as album art size? I think just because we won't be loading big themes that's no reason not to voice the other information. as again following your argument, people who rely on voice might not care about any of the information to do with skin buffer. I think both bits of information should be voiced just for completeness,as it seems crazy to voice one bit of info but not the other.
Re: Rockbox Presentation for New York Linux Users Group on 19 August 2009
snip I need to create a PowerPoint presentation (well, okay, not strictly PowerPoint; I'm using OpenDocument Presentation ;) ) to show as I talk, but I will also be using UI sims to show off the interface and show people how it works. Hi. Just a thought, but could you use forgive me I can't remember the spelling of his name, but could you use Tomers work on hid to control the presentation? that way you could actually demonstrate a live feature of rockbox, as I am sure he said one thing it could be used for was controlling power point presentations, this is just a thought of course.
Re: question about resyncing patch after wps code cleanup.
code cleanup. So if someone could please tell me what file I should be patching instead of gwps.c I would appreciate it. Look through each of the functions your patch changes, and find them in the current source code. For this you can look at the SVN commit logs or use grep -r to search for them in the source code. Hi thanks for your reply, luckily I think this patch is fairly simple, and actually all I had to do was change the patch file so that it patched /apps/music_screen.c probably not a very neat way to do things, as the file is patched with an offset of 266 lines, but seeing as this patch is only for my use, and all it does is insert functions I guess it doesn't really matter.
Re: fs #10336
or was your question what happens if you are on a file or folder name and release the joystick. well, in that event the name is spelled out. I don't have a problem with this, as if you are just scrolling by single up or down movements of the joystick you need to keep track of where you are. but if you are scrolling to the end or the beginning my argument is you should know where you are going to end up, and so it would be good if releasing the joystick cut the speech off. Yes, this is what I meant. If simply releasing the button were to truncate voice output, then it would do it even if you're only moving up and down one item at a time and that would make voice navigation useless. I suppose it could do this: a) Only if the button was held (i.e. you moved multiple items due to key repeat)? b) Or only if you're at the top/bottom of a list? c) Or only if both of the above apply? Hi. If I understand your suggestion right, I think c would probably be the best option. My idea would be that speech output is not cut off by releasing of the joystick, unless you are scrolling to the top or bottom of a list in the file browser, this would result in the file name or folder name being read, or in my case spelled which is why I think this feature would be useful. releasing of the joystick would result in the speech being stopped. Of course, scrolling single items and having the speech cut off when the joystick is released would not be desirable at all, even if you were reaching the top or bottom of a list. I'd prefer to keep it simple... ;-) Yes I agree, I like the sound of the idea in my own mind, but I have a feeling of course I would actually have to see it in practice to get a feeling if it was really desirable.
fs #10336
Hi. I am wondering, would it be possible for this to be resynced please? as it appears to have gone out of sync recently. I also tried posting a comment in the tracker with a suggestion for improvement, if this could be implemented, do you think it might be possible to commit fs #10336? the reason I am writing to the dev list is that My last comment on the task did not elicit a response, I am guessing pondlife didn't notice it or something. for reference, my comment with the suggestion for improvement in the tracker was However, I think what would be good would be if your browsing in the file browser particularly if you are using spelling of file or folder names would be that if you scroll to the beginning or end of a list by holding up or down, as soon as you release the joystick spelling of the file or folder name is interrupted. I think this would be good because otherwise you have to listen to a huge long name. Whereas if releasing the joystick cut the speech off it means you don't have to put up with this information. That is the only problem I have with this patch though. If my above idea could be introduced I would be quite happy to use this instead of Stephanes beep patch.
Re: fs #10336
Hi Alex, I'm probably not going to get time to look at this before the 9th of August, but I'd very much like to see it committed as long as it doesn't break anything. However, I think what would be good would be if your browsing in the file browser particularly if you are using spelling of file or folder names would be that if you scroll to the beginning or end of a list by holding up or down, as soon as you release the joystick spelling of the file or folder name is interrupted. I think this would be good because otherwise you have to listen to a huge long name. Whereas if releasing the joystick cut the speech off it means you don't have to put up with this information. If I understand you correctly, this should be a seperate patch. Personallly, I'd rather not interrupt the speech as I normally want to stop the list scrolling and then hear where I'm at in full. but presumably if your holding down the joystick to go to the end or beginning of a list of files or menus, you should already know where you want to be and what you will hear. the speech not interrupting isn't really an issue for menus, but I don't want to hear a massive folder name if I know that's where I was aiming for in the first place in the file browser, but of course mine is one oppinion, so I say let the majority rule.
Re: fs #10336
Hi Alex, OK, I've done a quick resync, but not tested the patch - please do so and report back on Flyspray. but presumably if your holding down the joystick to go to the end or beginning of a list of files or menus, you should already know where you want to be and what you will hear. the speech not interrupting isn't really an issue for menus, but I don't want to hear a massive folder name if I know that's where I was aiming for in the first place in the file browser, but of course mine is one oppinion, so I say let the majority rule. I see your point, but still think this should be a seperate patch. Hiow does the current (unpatched) code handle this situation? I am not arguing about the interruption of the speech being a separate patch. However in answer to your question, if you remember the hole point of fs# 10336 is that without it the speech does not announce an item when you scroll to the beginning or end of a list by holding up or down. 10336 simply makes the speech announce items at the end or beginning of a list if you scroll to it by holding up or down. or was your question what happens if you are on a file or folder name and release the joystick. well, in that event the name is spelled out. I don't have a problem with this, as if you are just scrolling by single up or down movements of the joystick you need to keep track of where you are. but if you are scrolling to the end or the beginning my argument is you should know where you are going to end up, and so it would be good if releasing the joystick cut the speech off.
Re: Run a build client!
Hi. Well I would be very happy to run an environment as a vmware image or something, however the rockbox images do not include any form of speech accessibility I guess to keep the size down. Wasn't there some guy a while back who was going to investigate the possibility of making an accessible vmware image for use with rockbox? If an accessible vmware image became available I would be quite happy to run it. Would a 1.80 ghz laptop with 512 mb of ram be any use to the system?
Re: Run a build client!
On Mon, 27 Jul 2009, alex wallis wrote: Would a 1.80 ghz laptop with 512 mb of ram be any use to the system? Yes it would indeed. We have several much slower machines already participating and they are doing good! OK, well I would be very happy to run a build client, but for the moment till either a version of the rockbox vmware image is made accessible or I can find an image with speech already added in I can't do anything. As unfortunately speech is needed in the image because a windows screen reader can't read anything in the image itself. So unfortunately I guess I need to find someone who can create a custom image for me with the speech installed. I don't feel I have the knowledge for this, because the most widely used screen reader for linux especially for command line work called speakup needs to be patched into a linux kernel, and I wouldn't have the first clue about how to start building my own kernel, or even how to put it into a linux distro let alone a vmware image.
status of fs #9067
Hi. I was wondering, does anyone know if anything is happening with fs#9067? I tried e mailing tom ross and leaving a comment in the tracker, but I guess maybe he doesn't watch the task or didn't get my e mail. The reason I ask, is that he said several months ago he was hoping to have free time to work on it and commit it. However the task itself hasn't been updated since march, and has gone out of sync. So I was just wondering if there is anything new I might be able to try out, or if it has been resynced, but not put on the tracker. Sorry for writing to the mailing list, but as I didn't get an e mail reply or a reply on the tracker task, I figured maybe the mailing list is the best way to attract mr. Rosses attention.
question about disc accesses
Hi list. I was wondering if anyone knows why I might be seeing the following behaviour. I have noticed that some times if I am listening to a dynamic playlist of around four hours give or take, when I get to the last track which would usually be around ten minutes in length, shortly before the end of the track literally a few minutes I get a disc spin up. Then once the track ends the disc spins back up in order to load my next folder. I was wondering if anyone has any thoughts why I might be getting this odd spin up shortly before the end of the last track. It doesn't always happen, but it does happen fairly often. I have the dircache option enabled, usually I get one disc spin up around half way through a playlist, and then some times this odd spin up shortly before its end. Wasn't fs #9621 going to try to deal with disc accesses? I did try having this patch applied on my build once, but unfortunately I found that whenever I wound back in a file my disc got spun up so I discarded that patch, but I guess maybe that's because it needs a bit of working on.
Re: question about disc accesses
2009/6/24 alex wallis alexwallis...@googlemail.com: Hi list. I was wondering if anyone knows why I might be seeing the following behaviour. I have noticed that some times if I am listening to a dynamic playlist of around four hours give or take, when I get to the last track which would usually be around ten minutes in length, shortly before the end of the track literally a few minutes I get a disc spin up. Then once the track ends the disc spins back up in order to load my next folder. I was wondering if anyone has any thoughts why I might be getting this odd spin up shortly before the end of the last track. It doesn't always happen, but it does happen fairly often. I have the dircache option enabled, usually I get one disc spin up around half way through a playlist, and then some times this odd spin up shortly before its end. Wasn't fs #9621 going to try to deal with disc accesses? I did try having this patch applied on my build once, but unfortunately I found that whenever I wound back in a file my disc got spun up so I discarded that patch, but I guess maybe that's because it needs a bit of working on. once again... this is not the correct mailing list... anyway, it could be spinning up for a number of reasons.. the obvious two are have you got it set to random folder advance? and its probably saving the playlist position or scrobbler or something Sorry for posting to the wrong list I didn't think. To answer your questions, I do not have the player set to random folder advance. I don't get why the player would be saving the playlist position if the file has say two or three minutes till it finishes, then it needs to start playing the next folder. What did you mean exactly, when you said the player might be saving the scrollbar?
Re: mp3 of devcon meetings?
I believe petur was looking at re-doing the mp3, since the one he posted earlier has sound coming from only the left channel - which is a bit odd. yeah, well, I'm a bit overloaded (and tired) as well, so the odd one is still up there. I'll have a look later on. No rush, I will wait and grab the propper version, after all I would probably only end up grabbing a corrected version anyway.
mp3 of devcon meetings?
Hi list. I have read the summary of what was decided at the recently concluded devcon, however I was just wondering if the mp3 of the meetings has been posted? as I looked at my messages carefully but couldn't find any reference apart from the one on daniels site saying he hoped to post it later. Thanks for any help. Alex
Re: Replaygain without a setting, and other menu cleaning.
cleaning. pondlife wrote: Sorry - I misread that. (I do think that the basic menu layout is a bigger usability issue than the points you've raised though - General settings indeed ;-) ) While the general layout is a problem, aside from replaygain the two screens I've mentioned are very hard to use on scrollwheel targets without a manual on hand and without advanced warning you won't know they're going to be so different and are unlikely to look them up in advance after learning the basic controls of Rockbox. Especially the pitch screen, which would require live experimentation to learn the controls. While the overall layout is a bigger problem, these problems are ones that a user will have a harder time coping with (I think) in the shorter term sense, because they aren't something that can be resolved by simple observation - they require actual experimentation, and in some cases may even trick a user into thinking something is impossible. With the menus, you can just iterate through all the lists until you see the option you want isn't there. With the pitch screen you may not even realize the screen can be toggled to different modes at all without the manual. So why not have some kind of menu when you first enter the pitch screen with the three modes to select from.
Re: Replaygain without a setting, and other menu cleaning.
alex wallis wrote: but if I understand your idea right, this would mean a user then has two things to alter if they want to change both pitch and speedh. I like having the flexibility of the three settings of just pitch, pitch and speed together and just speed. Removing one would create more work for a user if they want to change them together. How often do you want to increase the pitch exactly the same amount as the speed increase, so that the pitch is 30% higher AND the runtime of the file is 30% shorter? It's only for these very specific cases that there would be more for the user to do. Well actually, I use the option of being able to change both at the same time quite a bit when listening to music. Some times I just change the pitch without speed, though this is less frequent, but I do use all three modes. I just feel having all three modes gives a maximum level of flexibility to a user.
Re: jdgordon: r21307 - trunk/apps
Jonathan Gordon wrote: OK, nice patch... still not as clean as mine (I'm entitled to my opinion..) but if thats what it takes to be accepted so be it. I'll test it out tonight What about the subject of this thread? Will you be reverting r21307? Regards, Dave. My vote is to revert it as it just seems odd to have the menu change when there is nothing to resume. also I would have thought most people won't bother starting playback from the menu but will just hit play directly from the file browser once they turn there player on.
Re: jdgordon: r21307 - trunk/apps
That was Llorean's proposal to get around the frustrating splash which happens if you try to start playback when there is no playlist available, and there is no way of knowing that. I would still rather have done a dismissable splash, but whatever... 2009/6/16 pondlife pondl...@ntlworld.com: I agree, the previous wording made more sense. Could we not remove the menu item when it won't do anything useful? pondlife That would be fine too, however there is the whole argument against disappearing menu items, and having the order change for blind users. The change doesn't change the wording except when there is nothing there. It will still be Resume playback if its started up with a valid playlist. I don't see what's wrong with the old behaviour. Have I understood this right, that the behaviour has been changed so that if there is nothing to resume the resume playback option is renamed? is that right? I would have thought most people when they switch there players on are not going to go through the menu to the resume playback option, surely they will just press the play button or whatever it is called on there target which will start playback if there is anything to resume. Which would make this menu renaming when there is nothing to resume redundant and confusing. Because if they don't go through the menu to resume playback they won't know there is nothing to resume till they hit the play button meaning the menu renaming is pointless. Of course if there is nothing to resume a blind user will get a message saying this. So why exactly would someone go to the trouble of resuming playback from the menu? I don't particularly like the idea of disappearing menu items, I think it could easily confuse people and they could spend a lot of time looking for a disappeared item.
Re: Simpler solution?
pondlife wrote: As I see it, the problem is that list entries are never voiced until the button is released (and 0.25 of a second has passed). If list.c is modified so that first and last entries are voiced immediately, I don't think we'd need this beep, right? That's more or less what it sounds like to me. It seems like a bug to me that the last entry in either direction isn't voiced as long as the button is held down. Having the beep rather than the speech is more efficient than waiting for two or three letters to be read in a file name. A beep is not as intrusive as speech The beep you hear it and then its over but the speech will begin reading the hole file name which I would find annoying. Plus, imagine for whatever reason your finger slips from the joystick or wheel before you reach what you have been looking for. You will hear the speech starting to read, so some people if they were impatient might think they had reached what they were looking for. The beep is a nice safety measure against this possibility.
Re: Simpler solution?
alex wallis wrote: Having the beep rather than the speech is more efficient than waiting for two or three letters to be read in a file name. You don't need to wait for two or three letters to be read. The voice won't stop unless it's the last item in the list. A beep is not as intrusive as speech The beep you hear it and then its over but the speech will begin reading the hole file name which I would find annoying. The filename will still be read *after* the beep anyway, so I don't understand this point at all. That is not entirely correct, the file name is read after the beep if you are just going up or down with the joystick with single movements, however if you hold the joystick down in a given direction you just hear the beep. Admittedly it gets played continuously until you release the joystick, however this is an issue I guess that could be corrected. Why not give people this patch, but also a setting to enable it so that pondlifes speech option of having the first and last items read can be used as an alternative. After all, some people like beeps such as me, other people would probably prefer pondlifes solution if they are holding down the joystick or wheel to go up or down. Plus, imagine for whatever reason your finger slips from the joystick or wheel before you reach what you have been looking for. You will hear the speech starting to read, so some people if they were impatient might think they had reached what they were looking for. No, they won't, unless they're not thinking at all. There is such a thing as a mistake, and as I said what happens if you have multiple names that are nearly identical, the beep will flag instantly if they are on the beginning name or the second or third name as if they don't hear it they know they are at the second or third one. If they're looking for the end of the list, they'll simply press up once at the beginning of the list. If they're looking for anything other than the end of the list, they'll need to listen to what the list entry's name is anyway, to determine whether it's the item they're seeking or not. I don't agree with that argument, assume they know there item is three from the bottom, they scroll to the end of the list, knowing instantly they reached it when they hear the beep, they can then just go up rapidly a few times and bang they know they are on there item. Admittedly I guess the same would be true if pondlifes solution was used though. Seriously, half of your reasons for it seem to assume the person wants to get to the end of the list. If this is the case, it's significantly easier to just go up at the top than wait for a beep while scrolling. That assumes they are not down in the middle of a huge folder the beep indicates when either top or bottom is reached. The other half of your cases seem to assume the user is looking for something, in which case the end of the list is mostly irrelevant since that's not what they're looking for, and since scrolling will stop at the end of the list anyway, when they reach it, they'll hear that it's an item that's after the one they're looking for and start scrolling backward, which is something they'd have to do anyway. Instead of stating reasons why you think it's good, why not describe a normal use where you need the beep for the player to be usable - what would you be doing that requires knowing when you've specifically reached the end of the list (rather than any item that is after the one you're seeking), but where the end of the list is not your destination? The beep simply helps me keep track of my location in terms of where I am in a given list of files or folders. I did not know till now that holding the joystick in a given direction would allow me to stop at the end or beginning of the list. So I have been scrolling by rapidly moving up or down, and listening for that beep which told me instantly I had gone to the top or bottom of a list. But I don't really use it as a reference point when looking for a particular item.
Re: committal of patch
snip That's neat. Since the lists are circular, I would never have thought that holding the key down, which normally repeats, would stop at the list ends. It is inconsistent behaviour from a UI design point of view, but a handy feature to have. I never new about it either. If it is in the manual, I must have missed it. Incidentally, Alex, my trick for getting quickly to the beginning of a directory is to back up to the previous level (left arrow) and enter the directory again (right arrow). To get to the end, I do the same thing followed by a press of the up arrow. The two downsides of my method are that it involves a disk spin-up each time you do it and, of course, you can't do it on the root directory since the Files item on the main menu places you on the file that was highlighted the last time you were in the file browser. Why don't you enable the dircache option? I find for me at least it means I have a lot fewer disc spin ups, which I like as of course drives are increasingly hard to find these days. Cheers, Andrew.
Re: End of list beep
snip Incidentally, searching for a particular subdirectory/file buried in a directory somewhere is much easier on the iriver H1xx/3xx players with the LCD remote since one of the buttons allows you to skip through the list a bunch of tracks at a time. Could you please tell me how this is done with the remote? as the manual doesn't seem to contain descriptions of either the lcd or non lcd iriver remotes. I have opened a tracker task on this problem.
another suggestion for committal
Hi list. I have another suggestion for committal. That is the last two files I uploaded to fs #8006 I like the last version there specifically because not only does it give you a beep when the charger is removed or inserted into the player, but the beep_tune patch means that a short sequence of beeps is played so you can instantly know if you accidentally pulled the charger out or when its in that the player is charging. As otherwise you get no feedback at all to indicate that charging is working so you could easily find yourself trying to charge your player on a wall socket that isn't working and not realise this. But fs #8006 would tell you that the player wasn't charging if you didn't hear the beep.
Re: another suggestion for committal
Am Mo, 15.06.2009, 13:58, schrieb alex wallis: Hi list. I have another suggestion for committal. That is the last two files I uploaded to fs #8006 I like the last version there specifically because not only does it give you a beep when the charger is removed or inserted into the player, but the beep_tune patch means that a short sequence of beeps is played so you can instantly know if you accidentally pulled the charger out or when its in that the player is charging. As otherwise you get no feedback at all to indicate that charging is working so you could easily find yourself trying to charge your player on a wall socket that isn't working and not realise this. But fs #8006 would tell you that the player wasn't charging if you didn't hear the beep. We shouldn't overuse the beep. The more you use it, the less it means. Announcing pluging in or out could be very well done using voice. That could even differentiate between USB connection or charging only. Best regards. I agree about not over using the beep, however do you really need a huge speech message to tell you either the player is charging or the charger has been disconnected? the beep_tune patch allows a sequence of beeps differing in duration and length to play to give us an instant notification if the charger has been connected or disconnected.
Re: Advanced Solution
To add to the beep feature discussion in WM6 when you have lots of contacts and start scrolling through them for a few seconds you get a large letter appear on the screen which you then scroll through the alphabet rather than the contacts. This allows you to jump to the contact beginning with that letter very quickly. With speech and this feature in Rockbox it would allow much more efficient track/dir finding in Rockbox for both sighted and non sighted users. So have I understood your suggestion right, what you are suggesting is that if a person is scrolling through a large list of files by holding down up or down instead of moving through the list of files they get taken to a list of the alphabet, from which they then select a given letter and once selected are jumped to the start of the section in there list of files that begins with that letter? I hope my message is clear. The question is though how would you deal with numbers and non alphabet characters with such a feature? I think this feature would have to have some kind of setting as well.
Re: another suggestion for committal
alex wallis wrote: I agree about not over using the beep, however do you really need a huge speech message to tell you either the player is charging or the charger has been disconnected? the beep_tune patch allows a sequence of beeps differing in duration and length to play to give us an instant notification if the charger has been connected or disconnected. Someone who's never used Rockbox before will know what's happening when a voice says Charger Connected or USB Mode or Charger Disconnected or USB Disconnected. Meanwhile, a new user will have no clue what Beep beep bep means without having to reference a manual, something that may not be an option the first time they encounter it. Good point I hadn't considered that I guess I am used to using the version of fs #8006 that I use. I believe there is already a version of the patch up there that voices connection and disconnection information isn't there?
Re: Advanced Solution
This has been requested many times (in the same vein as being able to jump to the next start letter...) and yes its nice, its not really possible I have an idea how to do it but it wont look nearly as good as it does in WM... Why is it not really possible? Even if it doesn't look as good as WM, I would say that is far outweighed by the fact it would make navigation a lot more efficient. What's your idea anyway? maybe if you bounce it off the list it might spark a further idea in someone else. 2009/6/15 alex wallis alexwallis...@googlemail.com: To add to the beep feature discussion in WM6 when you have lots of contacts and start scrolling through them for a few seconds you get a large letter appear on the screen which you then scroll through the alphabet rather than the contacts. This allows you to jump to the contact beginning with that letter very quickly. With speech and this feature in Rockbox it would allow much more efficient track/dir finding in Rockbox for both sighted and non sighted users. So have I understood your suggestion right, what you are suggesting is that if a person is scrolling through a large list of files by holding down up or down instead of moving through the list of files they get taken to a list of the alphabet, from which they then select a given letter and once selected are jumped to the start of the section in there list of files that begins with that letter? I hope my message is clear. The question is though how would you deal with numbers and non alphabet characters with such a feature? I think this feature would have to have some kind of setting as well.
Re: another suggestion for committal
Barmy idea time! If you'd like beeps rather than spoken announcements, perhaps you could make your own voice file with beeps in...? You could have the whole lot in morse ;-) pondlife Lol that's quite funny. Unfortunately I don't no morse, but also I should imagine it would be very challenging to use. Its not I prefer beeps over voice, its just some times I think they would alert you a lot quicker to some information than a huge spoken announcement. But I take on board that beeps wouldn't mean a lot to the average user so perhaps spoken messages are better.
Re: committal of patch
2009/6/15 Dominik Riebeling dominik.riebel...@gmail.com We don't. We beep on manual track changes. Not on playlist end or automatic track changes. (At least not on my e200 with a current build and Track Skip Beep enabled.) I think Thomas means that we *should*, not that we do. If that is thomas's thinking, I don't really like the idea of a beep happening when an automatic track change happens, I would find it intrusive. I don't use beep when I do a manual track skip, because at least to my ears when I do it sounds like the beep is slightly distorting I don't know how to describe it. I don't know if that's a code issue or a minor player hardware issue. Though interestingly none of the other beeps my player produces sound kind of distorted as the track change one does. The level I have it set at doesn't make a difference to what I am noticing.
Re: committal of patch
snip They're simply two very different notifications, they just happen to both beep as their means of notifying you. That being said, I'm not really sure I understand why an end-of-list beep is necessary. If you're looking at the screen, it obviously isn't. If you're managing blind and using voice, you'll hear the start of the voice entry for the last thing in the list at the same time you would hear a beep if it were enabled. This is true, however navigating without this beep and simply relying on the speech for information to find out when you have reached either end of a list or menu makes navigation slower. Particularly in the case of files and folders, because just suppose for a second you wanted to know if you were at the end or beginning of the list, but at either end you had multiple entries beginning with the same letter or phrase, assuming you don't use talk clips which i don't, you will have to listen to the spelling of the name part way through to figure out are you on the exact end of the list, or maybe two or three items from one end. The beep simply gives you a very quick alert saying either you have gone to far and gone past what your looking for, or you have reached it. It simply saves a lot of time with navigation. If you didn't have this beep Given that you navigate lists to get to a specific option, if you're scrolling you'll need to listen to the voiced entries anyway to find the option you want. When do you need to know you're at the bottom, without needing to know what any of the list entries are, including the last one? see above, if its a massive list of files or folders particularly if they are almost identical in terms of there names. As a note: when navigating with voice, can't you already tell you're on the last entry by holding down until it stops changing entries? Since moving to the next entry should interrupt the voicing, as soon as it stops cutting off voicing and starts to read the whole entry, you'll know you're at the last one. I have found in the past that doing that can some times crash the player, but also this is not a very efficient way for a blind user to navigate, as if you are just holding down the joystick and looking for particular things you could easily miss them. at the moment, lists and menus wrap, so I am not sure wouldn't you just wrap around? also, just as a note, the beep comes in a while before the speech, there is a noticeable gap, making it more efficient than relying on the speech. What exactly is the intended purpose of this feature? to make navigation easier particularly in huge lists. I think though that perhaps this option needs its own setting, but I agree with previous posts from this thread perhaps beep settings need to be under some kind of category of there own. I think each form of beep should have its own setting though as not everyone would want all beeps.
Re: committal of patch
Thomas Martitiz wrote: I can imagine that it speeds up browsing for the blinds. And that's what Alex is reporting. Why can't you believe it? Best regards, I'm trying to understand exactly how this feature is used because so far he's only described it as it lets me identify the end of the list and I'd really like to know the use-case for spotting the end of the list since most lists you're looking for a specific entry in, and you should only ever reach the end if the last entry is the one you want. Hi Paul I understand you want to know why I find this patch useful. Well I actually didn't know that holding the joystick in a given direction continuously would take me to the end of a list and stop me there, as a side note, perhaps the manual should be updated to say this? however I just did a test, and once I was holding the joystick in a given direction to scroll through the menu, the speech wasn't playing, which I would find annoying if it was constantly trying to say things. The only reason I knew scrolling to the end of the main menu had finished was because I heard the beeping from the patch playing to indicate the end of the list had been reached. If the beeping hadn't been playing I wouldn't have known that scrolling had stopped and the end of the menu was reached. The same behaviour is also true in the file browser, so your information about holding the joystick and this patch should make browsing a lot quicker for me. I find the beep patch most useful though in the file browser as I have already said, but it is still nice to have it in menus. as for how long the beep happens before the speech, I haven't actually timed it, but I would estimate maybe half a second? you could always try the patch out yourself perhaps?
Re: committal of patch
alex wallis wrote: however I just did a test, and once I was holding the joystick in a given direction to scroll through the menu, the speech wasn't playing, which I would find annoying if it was constantly trying to say things. The only reason I knew scrolling to the end of the main menu had finished was because I heard the beeping from the patch playing to indicate the end of the list had been reached. So, if it started the voice for the last item in the list when the end of the list was reached, would your problem be solved without needing this patch? it would be solved but it would lead to slower browsing. as at the moment, I hear the beep from the patch and I know instantly I have reached the end of the menu or list. however, if the speech started to play for the last item in the list, I feel I would take longer to react to it as I would start to think about what letters or option I was being read. A beep is very short and easily attracts my attention something that speech probably wouldn't do. As I would take a few seconds to think about what was being read, but with the beep I know instantly what it means. Also, depending on the speed of the voice being used to make the voice file, it would probably take a lot longer to start reading whatever it was supposed to than a beep would take to play. Especially if it was a file name being spelled, as a beep is over like that, but by the time I react to the speech, the speech may have read two, three or four letters of a name.
question about committal of fs #8894
Hi just wondering when fs #8894 was committed earlier today was part of it missed out? as I am fairly sure that the last patch that was put up made changes to the manual in the pitch screen section to talk about the time stretch feature of adjusting speed without changing pitch. however, I can't see any of those changes at all. which makes me curious if either part of it was missed out or even if the wrong version of the patch was committed. as at the moment, the manual says It is not possible to change the pitch without changing the playback speed and vice versa. No mention is made at all of being able to adjust speed without changing pitch. As I understand it, three modes are given, the ability to change pitch and have the speed change accordingly, the ability to have pitch change without changing the speed, and the ability to just have speed change without affecting pitch. Pleas note I have been looking at the manual for h120/h140.
Re: question about committal of fs #8894
Hi just wondering when fs #8894 was committed earlier today was part of it missed out? as I am fairly sure that the last patch that was put up made changes to the manual in the pitch screen section to talk about the time stretch feature of adjusting speed without changing pitch. Alex, The manuals are only built once a day - so the change commited this morning won't be reflected in them yet. Download them again tomorrow morning however, and it should be there. Or you can build it yourself. Thanks for the information I did not know the manual was only built once a day.
Re: question about committal of fs #8894
2009/6/12 alex wallis alexwallis...@googlemail.com Thanks for the information I did not know the manual was only built once a day. The title says on the manual page Daily built manuals. Could we make it more obvious ? Lol I guess not, normally when I am browsing a page I am looking for particular information, guess I missed the title.
Re: a dev question about the Sansa Clip
Rafaël Carré wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Le Sat, 30 May 2009 15:40:00 -0500, Scott Berry scottbb1...@gmail.com a écrit : Hello there, At the moment I only have Windows. I do use Linux but do not have it set up for my Net book. Is there a way to do this through Windows? You can use cygwin environment, see http://www.rockbox.org/twiki/bin/view/Main/CygwinDevelopment If you use Linux, I advise you to use it rather than cygwin since cygwin can be very problematic. - -- Rafaël Carré I have had good success with Debian Linux running under Windows Vista 64-bit using VirtualBox. I don't know how well that would work on a netbook, and I do not know how easy it is for a blind person to set up, but it is Real Linux, and Real Windows. snip I am not really that used to linux, but I can speak from the perspective of a blind user. I should imagine in theory you could get a screen reader working in a virtual environment, however you would need sighted help, and as far as I know there are no sites that make virtual machines available with screen readers installed in them when you download them. This is just an observation not a criticism but I know rockbox does not provide a screen reader in the various virtual machines it makes available I guess because only a tiny minority of people would use it and also because it would increase the size of a package being made available. Unfortunately you would have to install a linux screen reader into a virtual machine no getting away from it because once you switch the virtual machine on, jaws will not be able to read you anything that the machine itself is doing, you have to have a linux screen reader for that. Personally I would love to get away from cygwin for my rockbox needs and switch to a virtual machine but it would be complicated to set up, and unfortunately people in my house are not that technical and probably do not have the patients to help me set up a proper virtual machine that is accessible I can use for working with rockbox. However if someone was to make one available or was interested in the possibility of trying to make one available I would jump at the chance to use it as I just find cygwin so slow to use, but unfortunately it seems to be the easiest option at the moment. If I could get an accessible virtual environment I would be interested in helping out by being a build server for example, but it needs an accessible virtual machine. I hope the above info has been of use. Alex.
Re: question about sleep timer code
2009/5/26 alex wallis alexwallis...@googlemail.com: Hi list. Today I decided to sit down and have a real serious go at getting fs#8630 working with current svn code. Luckily two of the files were fairly easy to fix, and I was able to work out where the code should go in the latest file versions. However, I am not at all sure about what needs to be done in /apps/menus/main_menu.c A lot of the code looks broadly the same, however I am finding the code a little bit hard to read, as there don't seem to be many comments in this section, and there seem to have been a few changes to main_menu.c since the last sync of the patch that I have which is for r. 18277. I would appreciate any advice. I can post the latest sync I do have to the tracker if that would help. There are just a few lines that have been added in since the last sync As far as I can tell, fs#8630 simply moves most of the code relating to the sleep timer into /apps/settings_list.c out of main_menu.c it also adds an int into settings.h the things that I am not sure about when I am resyncing are an if statement that relates to if a target has an rtc, if I need to stick that in /apps/settings_list.c as it is bundled in with the sleep timer code. There are also bits of code that relate to rounding up of values. So I would appreciate some advice. Let me no and I can stick the last sync of the patch up on the tracker. I hope it was ok to post this question to the list. Thanks for any help. Alex. yes this is the right list for it... I can't see anything difficult about syncing that, its all code which is just moved from one file to another (like you said), so just put it anywhere at the top of settings_list.c. I'm pretty sure that the sleep timer doesnt need the rtc, so make sure the code is all outside of any #if blocks. Hi thanks for your reply. I moved the code into settings_list.c from main_menu.c, and the code compiles. However I am having some problems that I would appreciate some help with. During the build process I get a message saying that sleep_timer is defined but not used. Also, when I run the build on target, the sleep_timer option is no longer voiced, and when I press select on it the player locks up and I have to hard reset, so guessing I have seriously broken something.
question about sleep timer code
Hi list. Today I decided to sit down and have a real serious go at getting fs#8630 working with current svn code. Luckily two of the files were fairly easy to fix, and I was able to work out where the code should go in the latest file versions. However, I am not at all sure about what needs to be done in /apps/menus/main_menu.c A lot of the code looks broadly the same, however I am finding the code a little bit hard to read, as there don't seem to be many comments in this section, and there seem to have been a few changes to main_menu.c since the last sync of the patch that I have which is for r. 18277. I would appreciate any advice. I can post the latest sync I do have to the tracker if that would help. There are just a few lines that have been added in since the last sync As far as I can tell, fs#8630 simply moves most of the code relating to the sleep timer into /apps/settings_list.c out of main_menu.c it also adds an int into settings.h the things that I am not sure about when I am resyncing are an if statement that relates to if a target has an rtc, if I need to stick that in /apps/settings_list.c as it is bundled in with the sleep timer code. There are also bits of code that relate to rounding up of values. So I would appreciate some advice. Let me no and I can stick the last sync of the patch up on the tracker. I hope it was ok to post this question to the list. Thanks for any help. Alex.
question about using svn
Hi. I have a question about using svn that I hope someone can help me with. How do I get a diff created showing me the differences between the local version of a file, and the latest version on svn? I tried the command svn diff talk.c from the apps folder, and my computer sat there for a bit before returning me to the prompt in cygwin. It didn't create a diff showing me the differences. Any help would be appreciated. Alex.
Re: question about using svn
Am 19.05.2009 22:10, schrieb Jonathan Gordon: http://google.com 2009/5/19 alex wallis alexwallis...@googlemail.com: Hi. I have a question about using svn that I hope someone can help me with. How do I get a diff created showing me the differences between the local version of a file, and the latest version on svn? I tried the command svn diff talk.c from the apps folder, and my computer sat there for a bit before returning me to the prompt in cygwin. It didn't create a diff showing me the differences. Any help would be appreciated. Alex. amazingly.. if there is nothing changed, it wont display any changes! He's missing the (as in: svn diff talk.c), that's why it's doing nothing. Thanks for your help, I tried the command you suggested, svn diff talk.c from the apps folder, and I am still getting nothing. Where exactly should the diff files be being placed? and am I right in thinking it would be called something like talk.c.diff? I looked in the apps folder but can't see anything like that. I no there are differences between my local and svn copies, as I have got r.20976 on my computer, and talk.c was updated in r.20977. The reason I want to know about the exact differences is because a few patches I run have stopped working so I want to see if its just a case of changing line numbers in them, or if I have to do something a bit more substantial.
Re: question about using svn
alex wallis wrote: I no there are differences between my local and svn copies, as I have got r.20976 on my computer, and talk.c was updated in r.20977. The reason I want to know about the exact differences is because a few patches I run have stopped working so I want to see if its just a case of changing line numbers in them, or if I have to do something a bit more substantial. You didn't mention that you were trying to diff between two SVN revisions. Most people were giving you help on how to diff between locally modified files and whatever revision your local copy was. Try svn diff -r 20976:20977 talk.c I think.** Hi thanks for your help, I got a bit more in the diff than I needed, as it gave me a full diff between r.20976 and r. 20977 but I still got the info I needed now I just need to figure out what to do with these stupid patches lol I am hoping its just going to be a case of changing line numbering slightly..
question about hid in rockbox
Hi. I know that one of the gsoc projects is to try to introduce hid to rockbox. I have read the hid entry on wikipedia, so I understand basically what it is, I was wondering is this going to make life for programmers easier, or will actual rockbox users see benefits from the introduction of hid? If so, what kind of benefits might they be?
question about USB connection of h140 to pc.
Hi. I have a question about the connection of an h140 to my pc. Before rockbox is put on a 140, when you connect it through USB to the computer the player automatically switches on and powers up. However when you install rockbox, you always have to switch the player on manually then connect it. So I was wondering why does this change in behaviour happen? is this automatic switch on a feature that just hasn't been put into rockbox yet?
Re: question about hid in rockbox
I have read the hid entry on wikipedia, so I understand basically what it is, I was wondering is this going to make life for programmers easier, or will actual rockbox users see benefits from the introduction of hid? If so, what kind of benefits might they be? Alex, my GSoC work is composed of two somewhat different parts. The first part is making the DAP enumerate as HID, which is in simple words making it enumerate to the operating system as a sort of multimedia keyboard. Today, the keypad has no use when the player is connected to the computer. With my work it will be possible to send the computer key presses, like controlling volume and skipping tracks. I am not trying to demean your work in any way, but if a user has there player connected to the computer at the same time as listening to music, why would they want to be able to use the player for adjusting volume and skipping tracks? as presumably they would be listening to music with winamp or a similar program which can be used to change volume and skip tracks anyway. In the future, it would be easier for developers to add any other keyboard mappings, such as using Rockbox as a Power-Point remote control. I can sort of see the benefits of this, although I guess you would have to be connected to a computer for that? This work is useful both for users and developers. snip It should make it easier for developers to create / merge additional class drivers to Rockbox, such as making Rockbox recognized as an MTP device (as some players in the market do). What exactly is an mtp device?
guidance with compiling rbutil under windows
Hi. I was wondering, could I have some guidance with compiling rockbox utility under windows? as I have decided its probably a good idea for me to be able to build my own binaries when patches for it are posted in order to test for accessibility. I have tried to follow the directions given in the rockbox utility development wiki, but I have almost certainly done something wrong or not done something I should have as the build keeps failing. I think really there should be some kind of guide to compiling rockbox utility that details exactly how to do it like there is for rb code. Anyway, what I have done is as follows. I downloaded and installed the qt sdk for windows, making sure to grab the version that said it came with mingw. I installed it, then I saw on the rockbox utility development page that the version of mingw that comes with the sdk is not sufficient for what we need, so then I went to the mingw site and pulled down the automated installer for the latest package. I selected full install, and to install it under the qt folder in the mingw folder. However I didn't delete any of the existing files that were there. Next I downloaded the windows api v3.11 as directed by the rockbox development wiki and put its files where it said they should go. After that I tried to build the utility, by loading the qt command prompt, going to the rbutil folder, typing qmake and then gmingw32-make after around half an hour the build failed. I am guessing I have probably missed out a lot of steps and not configured a lot of things, as there wasn't too much info on the development wiki page. In fact, I couldn't locate one thing it said I needed,called librspeex as it didn't provide a download link. so I would appreciate any help. Alex.
Re: report on accessibility problems of rockbox utility
Hi Alex, I didn't mean your question was OT, I meant my interjection! I was simply trying to say that once you can use RBUtil then the pool issue is a non-issue. If you could maybe have a sighted user configure the TTS options in RBUtil (a one-off thing), then it sounds like you'll be able to use it via the menu bar: Actions Accessibility Create Voice File, then ALT-I (for Install). The following keystrokes are equivalent: ALT-S, A, V, ALT-I. None of this should (of course) put you and Dominik off fixing up the Jaws issues, but I hope it's helpful - probably easier than mucking about with Cygwin, I'd think. Yes eventually I would like to switch to using rbutil for voice generation, however as well as the jaws issues, there is still that bug in qt which has been reported that means voices are not listed, meaning if I understand it right, that once I select sapi I can't select a particular voice I want to use for voice generation. Unless of course the utility will use the voice I have as my default system voice, which is a partial work around though not ideal as it would be good to select what sapi voice to use from within the utility.
Re: report on accessibility problems of rockbox utility
On Thu, May 14, 2009 at 4:58 PM, Dominik Wenger domon...@googlemail.com wrote: The current bug in rbutil prevents all sapi function from working. So no, unless this bug is fixed (or the windows binary is built with a earlier version if Qt) no voice generation with Sapi will work. The latest svn binary I provided is built with Qt 4.5.0 which didn't seem to be affected by the resources copy bug that caused the SAPI voices to cease working. When you say latest binary, did you post one today, including the patch that was posted a few hours ago? sorry I wasn't sure from your above message.
Re: report on accessibility problems of rockbox utility
alex wallis schrieb: Yes eventually I would like to switch to using rbutil for voice generation, however as well as the jaws issues, there is still that bug in qt which has been reported that means voices are not listed, meaning if I understand it right, that once I select sapi I can't select a particular voice I want to use for voice generation. Unless of course the utility will use the voice I have as my default system voice, which is a partial work around though not ideal as it would be good to select what sapi voice to use from within the utility. The current bug in rbutil prevents all sapi function from working. So no, unless this bug is fixed (or the windows binary is built with a earlier version if Qt) no voice generation with Sapi will work. But wouldn't building with an earlier version of qt result in a loss of accessibility features? with luck the bug will be fixed, I heard that trolltech is fairly helpful when it comes to accessibility.
Re: report on accessibility problems of rockbox utility
alex wallis schrieb: Firstly none of the boxes with number values that change when you press up or down arrows are announced when you make changes, you have to tab out of and back into them, so the same problem from 1.2.1 is still present. Ofcourse, we didnt change anything for that. But i am investigating the Problem. However as requested I pressed spacebar on the combo boxes and once they had expanded jaws is able to read items as they change. I actually didn't know that spacebar expanded combo boxes. However I think that the issue where combo box values are not read unless space is pressed on the box first should be fixed because I think a lot of people will just assume they can go up and down in the box without pressing space on it first. Also, once you have pressed space on a box and its items are being read by jaws, once you have found the item you are looking for you then can't do anything. As pressing tab will not move you out of the box, and if you press escape your selection is cancelled and you are back at square one Yes, surely the voicing of the comboboxes should be fixed. I was just asking you to try this, so i get a better chance to find the problem. Also if you expand a combobox with space, use enter to select the value :-) I will give that a try and let you know if it works. Also, spacebar on the combo boxes to expand them and get them read does not work on that one who's value is not read when it is changed even when you tab out of and back into it. The box I am talking about is the one next to the language selection box if you press tab once you are on it. Pressing spacebar on it doesn't seem to expand it and its values are not read if you change them. In fact you are given no indication of any changes when you press up or down arrow. I think, that you are in the Sapi Configuration dialog and the box which doesnt speak, is the Voice selection. And this Box is empty, because of the bug bluebrother mentioned earlier. And if its Empty, it is ofcourse correct to not speak anything. But are the Titles for this Boxes correctly spoken ? Do you mean the titles for the combo boxes? if so, no none of the titles for combo boxes are spoken. Under encoder settings as well as the boxes with number values not being read when they are changed, there is also a check box who's title is not announced, to find it move to the box who's default value is set at ten and press tab once, you are then on the box who's title isn't being announced. But the titles of the other boxes are readout ? Yes that's right, they appear to be read out correctly. Both the title on the check box for use sapi four and narrow band check box are read out. Also keep in mind, that the configuration dialog for TTS and Encoders change depending on which TTS/Encoder you selected. I see, I did not realise this. To you other mails: - The main menu contains both the manual and the info tab. But they are probably not very interesting for blind users. (Its the last two tabs, after the Uninstallation tab) When you say menu do you actually mean the menu bar with options such as accessibility and extras etc. In my mind I think of the menu bar as being separate from the tabs, but allowing you to do the same as you can normally on them. - Yes the voice generation in rbutil surely can be improved, as there are still missing features if you compare against the buildsystem. (POOL support for reusing clips, and voicestring correction for example) But as always we dont have enough time todo all we want :-) Lol yes never enough time in the day. Speaking of the build system and pool support, can you please tell me the exact syntax I need to use to get it working? as I am sure I am putting in the right commands, but cygwin doesn't seem to be using them, and is not placing the voice clips where I tell it. I would like to store the voice clips in a folder called voice located inside my build dir. If you could please tell me the exact command I should put into cygwin to do this I would appreciate it. Does cygwin need to have any particular packages or versions of packages installed for pool support to work? as I guess either I am doing something wrong or something isn't working properly with cygwin. I checked the entry on voice in the wiki, but couldn't see any information about pool support, I will edit the entry to include information on pool support if you can tell me how to get it working correctly. As I guess a lot of cygwin users might want to reuse voice clips.
Re: report on accessibility problems of rockbox utility
alex wallis schrieb: But are the Titles for this Boxes correctly spoken ? Do you mean the titles for the combo boxes? if so, no none of the titles for combo boxes are spoken. Could you please make sure again, that you are using the new version of rbutil (the link bluebrother gave you). Because the newer version shouldnt have a useSapi4 checkbox :-) Sorry, I have now used the version that bluebrother made available, I have now posted information in the tracker entry regarding combo boxes. I am just curious, why does the version of rockbox utility bluebrother made available not have a use sapi four check box? is support for sapi four being dropped? When you say menu do you actually mean the menu bar with options such as accessibility and extras etc. In my mind I think of the menu bar as being separate from the tabs, but allowing you to do the same as you can normally on them. I am sorry, with this i meant the main-screen, so not any Menu :-). If you are on the first Tab (Quickstart, the one with complete and small Installation) you press ctrl+tab 5 times, and get to the Tab with the Manual information. The tab after this, contains the Info screen. I was able to locate the manual screen using your directions, and it appears to be accessible with jaws. However, although I was able to download a manual for my player no problem, I could not open an on line version in my browser. No matter what I pressed, I couldn't click the link to open the on line manual. I think I also located the info tab, but when I navigated around it, all I was seeing was a blank field that jaws was not able to read anything from. However, I am not running a daily build on my player it is a custom one, would that make a difference to the info tab? I haven't posted in the tracker task about this problem yet. Speaking of the build system and pool support, can you please tell me the exact syntax I need to use to get it working? as I am sure I am putting in the right commands, but cygwin doesn't seem to be using them, and is not placing the voice clips where I tell it. To get the buildsystem to use a POOL of voiceclips, define a enviroment variable named POOL which points to the directory you want to store the clips. So in cygwin type: export POOL=voice before you type the make voice command. Many thanks for the information I think I no what I was doing wrong, I was putting a / before voice.
Re: problem with using pool command to store voice clips
voice clips alex wallis wrote: Hi. For some reason, I can't get the pool command to work with cygwin to store voice clips. I have tried using both pool=voice and pool=/voice You need to write POOL in all uppercase. Many thanks for this, I didn't realise the case was important, I will try again with the command.
Re: new binary of rockbox utility for me to test.
On Wed, May 13, 2009 at 5:51 PM, alex wallis alexwallis...@googlemail.com wrote: I see that Domonoky has posted a patch to fs #10205 that should deal with some accessibility problems. I am happy to test it, if someone could make available a binary including this patch. Binary of current svn with the latest patch of that task applied at the same location I posted yesterday: http://www.alice-dsl.net/dominik.riebeling/rockbox/rbutilqt-svn.zip Thanks for the binary, I am giving my feedback here, as its very detailed, and I get the feeling that huge comments are not encouraged on the tracker. So my feedback is as follows on the version with the patches applied from fs #10205 With the set of patches applied from this task, some problems have been solved, and new ones created. The biggest problem which has been created, is that the menu system is now completely inaccessible. If you press alt to access the menus, jaws is able to read the menu bar as you move along it items such as file and actions. However if you go into a menu, jaws will read you the first item, and then if you press up or down it won't read anything else, you get no indication of changing the menu selection. Several problems have been solved now, the titles on all tabs appear to be being read correctly, but there are still problems with the titles of some combo boxes and spin boxes. Which I will detail below. If you go into the option called create voice files, but don't click the button to change generation settings, the language combo box title is not read, you simply get the first language in the list read. The next problem we encounter is if we change the synthesizer being used to generate the voice. The title of the combo box listing them is read, and if you go up and down in it as your selection changes it is read. However if you then click the button to configure tts, jaws does not announce the title of the synthesizer you have selected to use, instead even if your selection says sapi, when you select configure tts, jaws reads the title of the dialogue as espeak configuration. However under tts configuration, the title of the language selection box is read correctly with jaws, and I am able to navigate it. Other controls in the configure tts dialogue are voiced correctly, these are the edit field called options, the voice combo box, and the speed spin box. All controls in the configure encoder dialogue seem to be voiced correctly, at least there titles are read. However, none of the spin boxes are read when you make changes to them, you have to still tab away and back to them. So the titles of the spin boxes are read, but not changes. There are still some things that are not being voiced, one tab title I can't understand at all with jaws, this is still under the generation settings change dialogue, the tab title I can't understand is located between the tts and encoder settings, and the tab called language. On the tab called language, I am seeing a single edit field, who's title is not being read. Back in the main create voice files dialogue, there is one spin box who's title is still not being read, to find it, if you place yourself on the cancel button and press tab once you will find it, its default value is 500. There is also one edit field who's title is not being read under the install talk files dialogue. To find it, place yourself on the cancel button, press tab once and you will find it, if you press tab again from it you find yourself on the brows button. Just as a general note, its not a good idea to rely on narrator for your accessibility tests, narrator functions very differently to a screen reader, and it is not intended to be one. There are a lot of things narrator can't read that jaws can, so narrator will not give you accurate test results. There is a jaws demo available that you could download and install on your system. I can get its url for you if you like.
Re: problem with using pool command to store voice clips
as a general rule... in the unix world (so cygwin also) case is always important... and even if it wasnt.. when someone types POOL=blaa it should be a hint that it might be important I appreciate what you say, and I will bare that in mind for the future. The pool command is now working. I don't get why they make the command case sensitive though, is a lower case pool used for something different to upper case POOL?
Re: new binary of rockbox utility for me to test.
alex wallis wrote: I get the feeling that huge comments are not encouraged on the tracker. May I ask where this impression comes from? The entire point of the tracker is for bug reports or patches and the comments on those reports and patches. These comments can be any size, and are welcome in any size necessary to convey the appropriate information. In fact, it's strongly encourage that comments go there so that future developers on that patch can see the train of thoughts and discussion leading up to its current state in a single place. If there's somewhere on the Rockbox website that suggests longer comments are unwelcome, please, let me know so I can see about clearing up any confusion around exactly what is said. There isn't a particular part of the website that says long comments are not allowed, Its just I have never really scene any monster sized comments on tasks before, and I didn't want to go posting something really huge. I was worried there might be some kind of unwritten rule, and I didn't fancy getting my throat jumped down so I thought I was playing safe by posting to the mailing list.
Re: report on accessibility problems of rockbox utility
On Mon, May 11, 2009 at 6:35 PM, alex wallis alexwallis...@googlemail.com wrote: OK, the version of the utility I am running is v 1.2.1 is this not compiled with the latest svn code? No. Rockbox Utility isn't build regularly as the main Rockbox itself is. Instead, we make releases every now and then. The rework Dominik talked about is in svn, but the changes haven't been made available in a release. I've just made a new binary available from svn. This is not a release and some internal things have been changed lately, so it may contain other issues. I'll try to update that binary every now and then. You can get it here: http://www.alice-dsl.net/dominik.riebeling/rockbox/rbutilqt-svn.zip Hi thanks for making the binary available. I have just downloaded it, and here are my observations. Firstly none of the boxes with number values that change when you press up or down arrows are announced when you make changes, you have to tab out of and back into them, so the same problem from 1.2.1 is still present. The same is still true of combo boxes when you tab to them if you simply hit down arrow the value changes but it is not spoken. However as requested I pressed spacebar on the combo boxes and once they had expanded jaws is able to read items as they change. I actually didn't know that spacebar expanded combo boxes. However I think that the issue where combo box values are not read unless space is pressed on the box first should be fixed because I think a lot of people will just assume they can go up and down in the box without pressing space on it first. Also, once you have pressed space on a box and its items are being read by jaws, once you have found the item you are looking for you then can't do anything. As pressing tab will not move you out of the box, and if you press escape your selection is cancelled and you are back at square one. Also, spacebar on the combo boxes to expand them and get them read does not work on that one who's value is not read when it is changed even when you tab out of and back into it. The box I am talking about is the one next to the language selection box if you press tab once you are on it. Pressing spacebar on it doesn't seem to expand it and its values are not read if you change them. In fact you are given no indication of any changes when you press up or down arrow. Under encoder settings as well as the boxes with number values not being read when they are changed, there is also a check box who's title is not announced, to find it move to the box who's default value is set at ten and press tab once, you are then on the box who's title isn't being announced. I have also looked at the about dialogue, and little seems to have changed in it, however now when I go into the about box, when I tab around the controls I am seeing an edit field with a list of names, but the title of this field is not read out, and of course just seeing that random list of names I have no idea what it means. Also, when you tab to the box containing the license it is not possible to scroll through it with the arrow keys meaning that if you want to read it you have to listen to the hole lot being read because when you hit down it will start reading from the top again.
feature suggestion for rockbox utility
Hi. I have an idea which I would like to see added to the feature request section if discussion can agree that it would be a useful feature. What I would like to see would be the ability for the rockbox utility to save voice clips that it uses to make a voice in a user specifiable folder for reuse next time a voice is generated. as often, a lot of clips don't change so it would save a lot of time if clips could be saved and reused when making a new voice, obviously generating new ones or replacing out of date ones where appropriate. I know it is possible to do this with cygwin, but I think this could be a useful feature for the utility as well. For example clips that don't change often are the menus and settings within them. It takes me around an hour and a half to make a voice for rockbox because firstly normally I use cygwin, but also I use a very high quality synthesizer, so if clips could be saved with the utility it would make a difference for me but also for the build process in general I am thinking.
can't find some items listed as being features of rbutil
Hi. I was just looking at the wiki entry for rockbox utility, and I was looking at the features of the utility that it lists. There are a few that I can't find. So I am wondering are these just accessibility issues, or are they not features of the rockbox utility? In the feature list the wiki says . Manual . Access to the Manual of your device . Info . Shows all installed parts of Rockbox I can't find either of the above options on the menus or tabs.
Re: report on accessibility problems of rockbox utility
Hi. Just to report, that rockbox installation on player is accessible with jaws, its possible to read the status of installation. Uninstallation is also accessible, I am able to set the options fine. I haven't tested install of voice file, but my feeling is that there is no reason why this won't be accessible with jaws. I haven't tested uninstalling of bootloader or installing of bootloader. Is this supported by the h140? as in doing it from the utility. I assume that you have to put the firmware file that a given operation needs on the player for the utility to be able to process it? so if your installing a patched firmware file, and if your uninstalling a clean file. I imagine the bootloader doesn't download and patch files for you or download clean files if your uninstalling.
accessibility of the rockbox utility
Hi list. I was just doing some research into qt, and I ran across a document detailing a lot of features that allow qt applications to work with technology such as screen readers. I guess implementing this would require a lot of code rewriting, especially as the utility isn't written in the latest qt. Am I right in thinking there isn't a tool to automate converting the code to work with the latest qt? anyway for those who might be interested the url I ran across is http://doc.trolltech.com/4.3/accessible.html
Re: accessibility of the rockbox utility
snip Well, maybe it would be best to start with the basics: how do you (or in general, how do blind people) expect it to work? Say, you downloaded this program (which doesn't need to be Rockbox Utility) and started it. You don't know the program, so how do you figure how to use it? Are you moving around with the mouse, trying to hover all buttons and stuff to get them read? Do you use the Tab key to go through all items you can select and expect them to be voiced? Do you use the Alt button to activate the menu and go through that with the arrow keys? Or something completely different? All your questions sound quite sensible, and I can fully appreciate you don't know how we do things. So to try and answer your questions. Blind people generally prefer a gui interface, as the majority of us use windows, and find it a lot easier to just select buttons in dialogues or go through menus rather than type into a command line which is a lot more fiddly and time consuming. The majority of us use the keyboard, because its a lot easier to do things than mess around with a mouse, though screen readers such as window eyes do have a feature where you can move your mouse and it will try to speak things as you move over them. But as I said most people prefer the keyboard. Generally when I start out with a new program the first thing I do is read its documentation to get an idea of how it all works, that way I have a basic understanding of the interface and know if I am going to be dealing with menus or tabs or in a lot of cases both. If I am navigating an interface with tabs I will be using tab to move between the various elements on a given page, and then if there is another tab I want to go to I will either hold down control and press tab, or if I am right on the part that will change the tab I will press right arrow. Either way when I press whatever keystroke I decide to change tabs with I will then expect jaws to announce the title of the new tab that I have just moved to. Its obviously quite important that titles of controls and tabs are announced as after a time I will know my way around the interface and be able to think where things are in relation to each other. If titles of controls and tabs are not announced, but we are just told that something is a button or a tab, then basically the application is inaccessible, as we won't have a clue what's what and what happens if we select a given button. We tend to activate buttons in dialogues or on tabs with enter or spacebar. Speaking for myself at this point I am sure people have other oppinions, but I don't tend to have a visual idea in my mind of what a given program is like, I usually have an idea after exploring the layout of the interface. To answer your question about the menus, I activate them with the alt key, I then expect to be able to move right or left along a menu bar, and then to be able to use the down arrow to open a given menu which will obviously contain items and maybe other sub menus. I guess the first start for me giving input is to do an accessibility comparison between jaws and nvda to examine the accessibility diferences. Yes, that would definitely be a good thing. Also, you should make sure to test with the latest version of Rockbox Utility, as there have been issues in Qt itself in the past. Generally speaking, newer versions of Rockbox Utility use newer versions of Qt, thus newer versions of Rockbox Utility should always be better from the accessibility point of view. I wouldn't be surprised if some issues are still around in Qt, but I definitely know that accessibility of Qt itself improved, which directly affects Rockbox Utility. I pulled down the compiled executable from the rockbox site. Is that the latest version I assume the utility is also built on a daily basis. Also, what exactly do you mean when you ask how a blind user sees the interface, the question is kind of broad. Are you asking how we navigate the user interfaces of applications? Yes, like this. My naive view on this topic is that I as sighted user have a visual impression of the application, so you need to have an impression of the application as well. For example, I consider the menu as a tree-like view: you have one top-level node File with the childs Install Rockbox Utility on player, Configure and Exit. Then there is another top-level node Actions with other childs. On the other hand, as sighted user I'm more driven by the visual appearance -- i.e. the buttons and tabs in the main interface rather than the menu. What impression does a blind user have? Is it more menu-centric? Are the tabs considered to have some specific structure? Generally speaking for myself I consider all programs being different, so I don't expect things to be in a particular order, and I am equally happy dealing with menus or tabs. From a layout point of view, I think the interface is fine, it just needs accessibility fixing, which I am happy to
report on accessibility problems of rockbox utility
Hi. As promised I have been having a look at the rockbox utility with jaws, and actually I must say since I last had a propper look at it a lot of accessibility problems seem to have been ironed out. However there are a few major ones that I can still see that I will now describe. These were all encountered with the latest version of rockbox utility, and the latest version of jaws. Firstly in the configuration dialog, the titles of the tabs are not read. So although I can read the contents of a given tab the title of it is not read to me when I go into it, jaws just says tab, meaning that I have to explore the controls to work out which tab I am on. In fact, throughout the utility the titles of none of the tabs are read. We have more serious accessibility problems though in the create voice file dialog. These are as follows firstly when you click the button to change the voice file generation settings, in the combo box that contains the lists of possible synthesizers that can be used for voice generation when you arrow up or down this box is not read. However if you hit the down arrow for example to change the synth to flight, and then tab to the next control in the dialog and then shift tab back this box is read. Obviously the correct behaviour for the screen reader should be that each time the synthesizer option is changed it should be read without having to tab out of and back into the combo box. The next problem is a bit harder to describe, outside the box to change voice generation settings there is a box who's default value is set at 500. There are two problems with this box, firstly its title is not read meaning that a blind user has no idea what this box controls. Secondly, if you arrow up and down within the box as the value changes jaws will not read the value to you as it is changed. However if you tab out of and back into the box it is read. The same problem is true of the box that controls the language of the generated voice, its title is not read, the only reason I know it controls the language of the voice to be generated is because the default value is at English. The contents of this box are not read when it is changed either so for example if you change from English to Spanish you wouldn't know you had done this till you tabbed out of the box and back into it. There is one combo box under configure tts that is not read at all, even if you tab out of and back into it. To locate it, if you are on the box to set the language, hit tab once and you are on the box that will not read at all, so obviously I have no idea what this particular box does. If you stay focused on this box that will not read, and then press your tab key once there is a box that has numbers you can change, its default value is 0, again jaws will not read the value as you change it, you have to tab away from and back to the box, also its title is not read meaning that we don't no what the box actually controls. Next to this box is a blank edit field, jaws just reads edit when I tab into it, so I have no idea what is supposed to go in there. The configure encoder settings suffer from similar problems. If you are focused on the cancel button and press tab you find a box with a default value of 1.0, jaws doesn't read the value as it changes, and also doesn't read its title so I don't no what its for. The same problem is true of the box that is next to the checkbox that says narrow band for its title, you press tab and find a box with a default value of 8.0, again, no indication from jaws of its title or what it might be for, and the value is not read as you change it with the up and down arrow keys, you can only find out what it is by tabbing away and back to the box. The same problem is true for the box next to that that has a default value of 10. Also, jaws is not able to read any information in either of the about boxes propperly, though it does read some kind of web address when you go to the option that is just called about, but it doesn't read any version information or anything. Also, the address is not read propperly. Interestingly, jaws is able to read the box that lists the different players that rockbox can be installed on with no problem. I hope my information has been of help, if you have any questions please feel free to ask, and let me no if you make changes and I will be happy to test them to see if they make any difference.
status of fs #9067
Hi. I was just wondering if there has been any improvements made to fs #9067. As I think that Tom Ross said he aimed to try to commit after the 3.2 release which I believe occurred fairly recently. I am looking forward to seeing this committed.
Re: Bring me your broken MP4s, your twitchy AACs and other non-working m4a's
Hi. I have a nice load of aac files that won't play with rockbox, they are all part of the same album, so I suspect I will probably only need to send you one file. So just wondering, do you want me to e mail it to you? or should I upload it to a site such as send space. the first track which is what I plan to send you is just over three mb. Of course if you would prefer I can upload the hole album, let me no what you want me to do.
Re: DevCon 2009
Unless somebody speaks up it will be June 19-21. Please update the wiki page if you intend to come or not, and if you plan to stay in the hotel I arranged. All info about this can also be found in the wiki page. Hi I am considering trying to make it to this years event, but have a few questions. I am not a programmer though trying to learn. I was wondering do people think there might be value in a non programmer attending? not sure how much use I could be to discussions. Also, would people have any objections to a blind rb user attending? have any of the blind rb users ever attended before? Thanks for your help. Alex.
Re: question about changes to the content on the rockbox homepage
alex wallis wrote: Hi list. I was just keeping my usual eye on the rockbox site today and noticed a few things I am wondering if someone could please explain. Firstly in the news section, what exactly does 2009-04-01 After months of negotiations, Apple has finally agreed to our terms and Rockbox is now an enthusiasic part of Apple's development team! mean? There is no link to get more information on what this means for rockbox and what exactly this partnership relates to. My second question is why has svn access been disabled? I wasn't planning to do a checkout today, but all it says on the page next to it is see news and irc. I didn't see anything in the news section about why svn access has been disabled. Just curious that's all, so hope someone can clear these issues up for me. I assure you, new information will be out tomorrow absolutely clearing up both of these issues. I see, but tell me does the svn access have anything to do with the new apple partnership? I hope apple isn't going to start being able to dictate the course of the project and what is and is not going to be included such as support or lack of for various formats etc.
Re: question about changes to the content on the rockbox homepage
Are you in a timezone where it is not yet midday? I don't understand your question. I am currently digging through the irc logs, and what I am reading is beginning to disturb me I really really hope rockbox hasn't sold out to apple and is not restricting development to a few hand picked people. I also really hope the project isn't going closed source and that we will still be able to build from source with patches created by ordinary users, and we won't be forced to have restrictions imposed on us by apple. I will be very disappointed if the apple hardware is the only players being developed for. I think its a very bad move posting news items like this and disabling svn access with no announcements as to the reason why. I think the people responsible should have waited till they were completely free to give explanations and posted announcements and explanations together.
Re: question about changes to the content on the rockbox homepage
Noone sold out... We merely came to a mututal agreement with the fine people at Apple to discontinue support for all other targets and work exclusivly on ipods and their upcoming devices... The great news here (other than the wonderful cash bonus most of us will get :D ) is the coming support for the entire ipod product range (including the iphone!).. but we have to keep this hush for now... You say you have not sold us out? your dropping support for all other devices to concentrate exclusively on apple products, I think that is really narrowing the projects focus. I see this as a betrayal of the projects principles and selling us out. Your abandoning all other rockbox users. I hope someone sets up an alternative project to rockbox. I think the community should have been consulted on the hole idea of doing this. I think the way this has just been announced as a done deal is extremely under handed and I can't believe this has been allowed to happen, and i think the phrase cash bonus says it all.
concerned about proposal to intigrate tts into rb core
Hi list. I haven't really been following the gsoc 2009 ideas, however having scene a few mails about the idea of intigrating tts into the rb core I felt I should write to the list and express a few concerns. For anyone who hasn't read the page with the proposal it is located at http://www.rockbox.org/twiki/bin/view/Main/SummerOfCode2009 I have no problem with the idea in principle with integrating tts into the core, I think it is a really good idea, I can see how it could solve a lot of accessibility headakes and some things not being voiced. However my main concern is if this is accepted, completed and intigrated into rockbox, would this be used as a way to drop support for the old voices completely, in other words we would have no more lang files, we wouldn't be able to generate our own voices and we would be forced to use the intigrated tts. My reason for asking this is that I could see myself using such an intigrated tts maybe for file browsing, but I wouldn't choose to use it routinely as I find the espeak voice horrible to use and very hard to understand, I don't no anyone who actually chooses to use this voice in day to day computer use, it would be a bit like someone using jaws with a default Microsoft voice. I prefer to generate my own voices currently for rockbox as I use very high quality human sounding synths to make my voices. If the two systems will be allowed to run along side each other then I am all in favour of intigrating tts into rockbox. My other reason for being concerned about the espeak tts is that I have to concentrate to understand it when I have to listen to it, but what about people who are hard of hearing? how will they manage with such a tts. These are just my thoughts. Alex.
question about location of information in lang files
Hi list. I have a question about location of information in lang files that I hope someone can help me with. What I would like to know is does it matter in a lang file where information is put, provided the rules are followed with regards to formatting and conventions of structure. For example, if a patch wants to put something after line 100 but the patch is out of date and it can't be fitted there but the phrase it wants to put in is still needed and the rest of the patch applied OK, would it be ok to just stick it at the end of the lang file. My reason for asking this is that I would like to test out the work that Tom Ross has been doing in FS #9067. However I am making a custom build and also including a load of patches made by Sdoyon. But because of the major changes made buy fs #9067 most of the lang file patches he created won't apply, even though the other bits of most of his patches will apply. So provided things don't have to be stuck in particular areas of the lang file what I am planning to do is just copy and paste the various things that his patches need at the end of the file. Even if that makes it look untidy its only for my own personal use. Thanks very much for any help. Alex.
Re: question about location of information in lang files
The order of the phrases in the lang files do not matter. Genlang will sort them automatically and the strings will be loaded to the correct array index by lang_load(). Thanks for that information. Regarding FS #9067, make sure you are using the latest patch posted in the task (sorted_genlang.2.diff). Most of the lang files are unchanged from SVN. The only lang files that are modified are english.lang and espanol.lang. That's fine, as I use English. Just wondering, at the moment using the latest diff from fs #9067 which plugins are currently voiced? From what I can tell it seems to be blackjack, chessbox and chopper is that about it? I was wondering how do you go about working out what strings for a given plugin need to go in the lang file? as I might give some of the others a go at least for english.lang I was just wondering with the chessbox plugin when a game is being played will the voice give you feedback when you move around the board for example if you are on an empty square or on a square with a peace what peace it is?
Re: question about location of information in lang files
lang files The voice changes in FS #9067 are more of a proof of concept at this point. Splashes and menus should be automatically voiced due to the ID2P() macro. Blackjack also includes some explicit voice changes and in theory you could probably play a hand or two without using the screen. The actual voicing is just roughly done for Blackjack and so it is fairly confusing. I haven't touched any other plugins at this point other than Blackjack, Chopper, and Chessbox. Basically all strings in a plugin will be moved to the lang files unless it makes sense for them to be hardcoded. Blackjack's start page is untranslated because I plan on reworking it with a menu at some point. I see, do you think it might be possible for me to play chess without having to look at the screen? I would be very interested in playing solitaire, what would be the best way to make a start translating its strings? if I managed to get anywhere I would upload a patch to fs #9067 as guess any strings translated into the lang file would help when it comes to putting them into there own files. Am I right in thinking chopper is a graphical game that a person would need site in order to play?
Re: how is strnatcmp aka Interpret numbers while sorting supposedto sort?
Maybe we should just see if various FLOSS file browsers have a common natural sort, and use it, so that our files are likely to show up in the host's browser the same order as they show up in ours? Now this idea I like, personally I think sorting should work the same way it does in windows. I guess if people don't like the way files are sorted they can always rename them.
Re: how is strnatcmp aka Interpret numbers while sorting supposedtosort?
alex wallis wrote: Now this idea I like, personally I think sorting should work the same way it does in windows. Windows isn't FLOSS and I very much doubt their algorithm is public. I see, I didn't no this, just out of interest what is floss exactly? I guess if people don't like the way files are sorted they can always rename them. No, actually they can't if the sorting tries to be too intelligent. By this argument, we should just keep our original sort and turn off all others, since it at least will never try to guess anything, and strictly use the filenames. It's the only way for renaming to be predictable and consistent. I think rockbox should include some sort of sorting, and not literally use the file names. Before 8314 it used to drive me mad that sorting went 1 10 etc. I really like the fact numbers are now sorted in the correct order. I can't really enter into the sorting debate very much as I hate leading 0s in file names, I like my file names to mostly be simple either being labled 1, 2,3 or 01, 02, 03. if file names include .s I like 1.1 1.2 1.3 etc. I make a point of removing leading 0s from names.
Re: FS#9873 - Utilise buttons for playlisting
I am more in favour of setting up a mechanism where spare buttons could be mapped flexibly to much more than just adding tracks to playlists..., jump to fm/rec/some menu item/etc Hi. Now please correct me if I am wrong, but what you wrote above sounds a bit like having the ability to configure buttons through different options. I thought that configurable buttons was something the rb project was dead against, although I disagree with this policy but anyway just thought I would throw my points in.
reassembling h140
Hi list. I am sorry for writing on this topic. However I am having real problems getting my h140 reassembled. I was wondering is there anyone in the UK who knows about assembling of h140s who might be able to physically help? I wasn't able to do any of the physical disassembly myself as I am blind, so I got my dad to help. the problem I am having is that my original player developed several problems, the joystick was beginning to go, and the remote wasn't functioning properly even on the original firmware. I tried with several remotes, however the same problem seemed to be happening with all three, so I knew it was a problem with the players boards. So I got a new player from ebay. I decided to install my old 2200 MA battery and my old hard drive into the new player from ebay. My old player disassembled fine using the guides from misticriver.net However the player I got from ebay was a nightmare, I believe that the original person who sold it to me must have got it second hand himself from someone who tried to do stuff to it but didn't no what they were doing as the threads on several screws were gone. However eventually the player was disassembled. But then my dad found for some reason he wasn't able to install the battery into the player from ebay, and he can't seem to get it to go back into the original player it came from. So at the moment both players are scattered over the table in bits, and we don't have a clue what to do. If there is someone in the UK who might be able to help that I could either send the players to or deliver them in person I would be very grateful for any help. As I am now without an mp3 player. I have asked on misticriver.net for some help, but thought I would ask on this list as well. I decided to post to this list as I thought this list would probably have more people on it who know about the hardware side of stuff than the ordinary users list. Any advice would be very welcome. Alex.