. The
few themes we have which include skinned lists can disable it if they wish,
just as they can modify any setting, no?
pondlife
.
pondlife
p.s. I apologise, sincerely**.
(*Ironic use of entirely inappropriate personal abuse, for which I should
apologise.)
(**Non-ironic; I really don't like this personal abuse thing - I'm sure
there are dedicated MLs elsewhere for that sort of thing.)
TBH, I would like to revert this commit.
FWIW, I agree. A solution is needed to this memory waste, but this isn't
it.
pondlife
Slightly OT - any chance this could somehow allow the user to choose native
lists on RaaA? It's something the Android build could sorely do with IMHO.
pondlife
-widget, with theming/viewports used to glue it
all nicely?
pondlife
and now can build ok. Will attempt a GCC 4 update soon though
- a better solution if I can get it working.
Thanks,
pondlife
: error: (Each undeclared
identifier is reported only once
/home/Steve/rockbox/uisimulator/common/io.c:350: error: for each function it
appears in.)
make: *** [/home/Steve/rockbox/buildsim/uisimulator/common/io.o] Error 1
Is this just me, or an actual bug?
pondlife
/rockbox/buildsim/apps/codecs/libspeex-voice/bits.o]
Error 1
Guess I need a newer gcc - am running 3.4.4 (cygming special, gdc 0.12,
using dmd 0.125). Last time I tried to get gcc4 working ujnder Cygwin (a
couple of months ago) there were plenty of problems; guess I should try
again?
pondlife
got to such a Windows-unfriendly conclusion -
both points 1 and 3).
pondlife
is broken and fixed that too.
That's great news - I'll contact him. (Don't think I'll be resuming the
bootloader work though...)
pondlife
Gets my vote. I'm in favour of the any button approach, although would
prefer the option to charge by default (i.e. button press to connect) as per
FS#10198.
pondlife
It's works now right? It seemed best to just abandon the convenience for
now and implement codec_main explicitly even if it doesn't have to do
anything.
Yes - thanks very much!
pondlife
under Cygwin, so this may
kill off Cygwin as an option for Windows users.
pondlife
misset my include paths (stdio.h is present in
both /usr/include and /usr/include/sys) but I can't see where configure
ought to be setting this. Is there some environment variable I'm missing?
Any clue wouild be very welcome.
pondlife
added per-file resume?
pondlife
from. If I want to requery to get
all-tracks-by-an-artist-including-that-one-I-just-added, then that's an easy
enough thing to do.
pondlife
without re-querying? IMHO
that's a fairly rare case not worth the code and confusion.
There ought to be as few differences between directory/database/M3U
playlists as possible (none, ideally).
pondlife
further.
pondlife
the previous playlist, thus
continuing with the correct program with no need for any clever code. (I
don't much like clever code.)
The thing you'd lose would be the automatic addition of new files, which
could be accomplished with a Refresh Playlist menu option.
pondlife
be an achievable GSOC project in itself, as well as
being one input for a larger database redesign project.
pondlife
Please could somebody who knows about system messages and/or queuing take a
quick look at http://www.rockbox.org/tracker/task/11927 ? This patch/hack
makes the sim usable for me again (since the introduction of
KERNEL_OBJECT_CHECK) and might help in a wider sense.
pondlife
,
is pointless IMHO, and takes a lot of the collaborative fun out of the
process too!
So - pop it on Flyspray, not in SVN..?
pondlife
to .mp3 and playing them?
pondlife
would need to be added to make sure it doesnt blow up
if the AAC codec gets a MP3 instead?
I'd hope in most places it shouldn't need explicit error checking, just
correctly-written code. ;-)
pondlife
Probably gettnig a bit OT - but what's the status of http://new.rockbox.org/
?
pondlife
to cover any database results that are a list
of albums or tracks. e.g. Database A to Z Album G (Pictureflow)
might be nice for speed of access.
What do you think?
pondlife
be confusing to have one thing in
multiple places in the UI. I'd rather aim to get the database integration
improved and stick it in there for now.
Assuming we're ok with customisable menus, a root menu entry straight into
the plugin may be achieved that way...
pondlife
Assuming we're ok with customisable menus, a root menu entry straight
into
the plugin may be achieved that way...
I'm so used to that being a nodo that I forgot that would be the
obvious answer :p
I suggest you implement it quickly before it becomes a nodo again ;)
pondlife
So will this port fly, or go belly up?
pondlife
I'd like to see this committed.
pondlife
seeing the colour/readability on target. A reboot, or
text file editing, should not be part of this)
Simple as that.
pondlife
really change themes?
Maybe with this there would be no reason to have the S/M/L/XL categories at
all, just work out the number of bytes required, compare to the number
allocated, and prompt for a reboot to change theme if needed?
pondlife
and grab the audio buffer? Loading a theme will need a spinup
anyway, so a rebuffer isn't too expensive...
pondlife
text file is loaded into the plugin buffer already).
That's the sort of thing I was thinking of. The RAM usage doesn't have to
be exact, as long as we're quite close and always round up...
pondlife
I dont think we should ship any themes at all apart from cabbie.
I agree.
So do I, FWIW
pondlife
targets
at the moment, so am restricted to the sim.
pondlife
?
Not the adapter, just a way to avoid soldering.
I really ought to get BDM kit sorted if I'm to fix up the bootloader, really
don't want to pester you every day or 2 ;)
pondlife
soldering, I'd definitely be up for that.
pondlife
model or mkboot is unable to cope with an image greater than 1k or what else
it might be. I implore anyone who understands map filws to check out my
recent post which had details attached!
pondlife
this will have to be somebody who has both an H300 and a BDM
debugger. I don't know if this includes anyone except Linus - but if it
does, now would be a good time for them to speak up.
pondlife
lend
the device to a dev).
I fear there are relatively few H300 owners around now, although it remains
my player of choice, so this might be a last chance to bring the bootloader
up to speed.
Does this sound like a reasonable idea, or a waste of time? Any other
suggestions?
pondlife
, but until then I fear they will gather dust. (In fact, I stopped
working on the RTC patch for this very reason.)
pondlife
favourite annoyance with the h300 bootloader is that it starts the
OF when connecting the charger while it's off, is this fixed in svn?
Yup, that's fixed. I'd forgotten about that - and I run off the charger a
lot of the time.
pondlife
or similar (assuming we keep it in...)
Hope this helps,
pondlife
Looking at the original crossfade logic, it would act as you describe
using Shuffle and Manual Skip, but the manual seemed to describe it
differently. So I used the AND logic when rewriting the decision tree.
I'll fix it back to an OR decision and rewrite the manual to make it clear
For what little it's worth, I'd also vote for just using %St. it's:
- More readable (IMHO!)
- Non-redundant (redundancy is redundancy, user side or not)
- Less stuff to document (less CustomWPS is less confusing for WPS authors)
- Less stuff to test and maintain
pondlife
should be tied to customisable
keymaps - I'm slightly against that idea, largely on the basis that if
there's a *better* keymap for a particular target then it should be coded
for all to enjoy.
pondlife
than
have two seperate methods for sighted and non-sighted use, I'd propose a
voice UI for it so we have a single model. A screen reader, if you will...
pondlife
.
Ideally it could be combined with an enhanced quick screen which would allow
you to attach a manu function (not just a setting) to one of the actions -
so you could then get the time/battery level/lots of other things by holding
down LONG MENU then UP (for example).
pondlife
Info -
which is where we came in...!
pondlife
is entirely configurable, you could
set up just the fields you want to hear - e.g. time, battery level, track
name, track progress.
Might be overkill, but it would bring the visual and voice UIs closer
together.
pondlife
the feature. The downside is that an explicit enable setting is required
and that a a reboot is needed after this is enabled.
pondlife
, and no
rebuffering).
pondlife
IMHO, it would be useful, but 7k sounds a bit OTT..
Does set_limiter_buffer() need to be inlined?
pondlife
as and when
new space was needed, using the hints given via the file type.
However, it seems to have become malloc-by-proxy. :)
pondlife
wrong buffer_alloc(), I'm using the one which steals some
audio_buffer before playback starts...
/me is slightly happier ;)
pondlife
question - why do we have the double zero? I'd prefer
config_1, config_2,... (or eq_1, eq_2,...) personally.
pondlife
pondlife, I may be wrong but I think adding the leading zero to
numbers 10 is standard convention? A bit like how music
tracks are always numbered track01, track02...
Fair enough, although that convention makes most sense for tracks ripped
from CD where 99 is the max track number
Rafael - good luck, we always need more playback (and buffering) gurus
amongst us!
pondlife
, then shutdown on
idle timeout and restart on alarm it ought definitely restart the playlist
(rather than silently display Nothing to resume). Consistent!
pondlife
be consistent and (particularly) for use as an alarm clock. But I
might have missed an earlier message giving a case where the opposite
behaviour would be desirable, feel free to restate such cases for my benefit
;-)
pondlife
). And a more reliable alarm clock!
pondlife
hamper development for the rest of us.
Could the iPod stop action be delayed so it only happens on key release
(i.e. when it's known to be a long press, not a very-long one)?
pondlife
the most logical to me.
Fair enough!
pondlife
I'd say that anything related to a particular patch (including non-technical
discussion) should be on FS , but more general stuff should be on the ML.
pondlife
We did indeed. You'll have to try and make it next year, Godeater has
volunteered to organise it in London :)
Excellent - I'll have almost no excuse then!
pondlife
not actually described why the screen is needed beyond The
nudge functionality should be preserved. The screen and the nudge
functionality aren't necessarily teh same thing.
Because it's a good UI and one of the best parts of Rockbox, IMHO (since way
back when in the Archos days).
pondlife
the up/down/left/right pitch screen.)
pondlife
to change pitch without changing BPM/perceived speed (i.e.
where speed and pitch change together)
Plus, the pitch/speed values are not settings - they are not persisted.
This, along with the above, is exactly what I want when I'm DJing.
pondlife
Hi Paul,
Also, if we're to address the menu layout (and we should), I'd suggest that
http://www.rockbox.org/twiki/bin/view/Main/MenuLayoutDiscussion offers the
best proposal. I know it needs a bit of bringing up to date, but Marc's
basic idea is good, IMHO.
pondlife
timestretch wouldn't be acceptable
for committing if it gave a compulsory RAM loss).
pondlife
- i.e. we should
have pitch on the wheel and speed/nudge on buttons?
pondlife
!)
In summmary, you've identified three improvements...
1) Enable timestretch by default
2) Display speed based on user's perception, not on algorithm parameter.
3) Swap pitchscreen axes on wheel targets
... but I still think the screen is needed.
pondlife
Hi David,
Take a look at http://www.rockbox.org/tracker/task/10341 - I'd be interested
to know if the patch there helps.
pondlife
I was thinking that the splash shoudl be dismissed by any button press, and
that button press should NOT be eaten - i.e. the splash is displayed (with
or without timeout), but you can carry on operating as normal.
pondlife
it
confusing at all - it actually removes any timing dependency and brings the
same model to the visual UI and voice UI.
3) the splash could be called in any screen which means the screen's
action context needs to be passed in...
Why?
pondlife
The problem on iPods is that there's something mapped to long Play and
something mapped to very long play, not the difference between long and
short.
So the correct fix would be to somehow scrap the very long concept, I
guess. (Not to introduce more options...)
pondlife
The patch on FS#10322 now ignores SYS_ events.
It should probably also ignore wheel events - could someone with a
scrollwheel give it a test run?
pondlife
it (or want to use it), what are the
actual outstanding issues?
pondlife
I agree, the previous wording made more sense.
Could we not remove the menu item when it won't do anything useful?
pondlife
by any key press?
pondlife
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?
pondlife
Good work!
At this point, I'd suggest that you host your patches on Flyspray as that
might be higher profile and help you get help (testing and/or developing)
for the buffering side of things.
pondlife
Either way, I think the list should speak the last entry before the button
is released.
I've put a patch up at FS#10336 and would welcome feedback on what it
breaks!
pondlife
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
Release is soon. Are we going to disable it again for 3.3 if we don't find
a solution soonish?
I vote to leave the code as it is. The algorithm is disabled by default
anyway, right?
A bit OT - What's the status of USB for 3.3?
pondlife
mucking about with
Cygwin, I'd think.
pondlife
find that generating voice files (and .talk files) is much faster in
RBUtil than with Cygwin.
IMHO there's no need to worry about pooling when using RBUtil - it's just
that Cygwin has really poor performance.
Yours,
pondlife
Hi Peter,
Sadly I won't be able to attend this year. Maybe in 2010 (when maybe I'll
have had more time to contribute to Rockbox in general).
pondlife
in the followup, was to remove
the setting as a whole.)
The third option was not to remove the setting, but to change it from
Prevent track skipping to Prevent skipping. Hope that's clearer now - I
shouldn't attempt to interleave paid work (TM) with Rockbox fun...
pondlife
for
seeking. IIUC, this was originally put into study mode to prevent
accidental track skips when seeking, but I might be wrong on this one.
pondlife
appears to be a track skip to the user.
pondlife
In this case we have three options:
But I obviously forgot the third option... which was probably going to be
- Disable skipping completely, not just Track skipping
I don't like this option any more. ;-)
pondlife
This worked, thanks. I take it this is more of a one-off that I don't need
to incorporate into my overnight build script?
I think so. lang.h is a bit special.
pondlife
Hi Mike,
A file was moved (I think it was lang.h, from memory) a while back. You
might need to delete the old copy.
pondlife
widespread benefit for
the project as a whole.
pondlife
Let me just share the info that we just now reached 100 builds in the
build table visible at: http://build.rockbox.org/dev.cgi
Congratulations! But wde... ;-)
pondlife
for all themes...?
pondlife
to it), then the patch needs a config
option anyway IMHO.
I was mainly pointing out that as long as the core (6) buttons aren't
configurable (which they definitely shouldn't be), what's the harm? As long
as we have a line in the sand, and a clear policy, that is.
pondlife
p.s. IMO
a USER
button that *is* configurable, defaulting to the following:
USER - do nothing
LONG USER - record
Hopefully such a simplistic view would be generally usable on all targets
and would help lead to consistency across the UI and across targets.
Yours,
Steve Bavin (aka pondlife)
1 - 100 of 169 matches
Mail list logo