Re: What's wrong with folder browsing?
Hi, ext Frantisek Dufka wrote: >>> You can also do MD5 sum of directory listing (even output of ls -R or >>> -lR) and do not scan files at all if listing hash is same. >> this is easy, but not much faster than the stats... > > Sure if you stat directories recursively it is same. The idea was just > to avoid reading the file data as much as possible. At least > metalayerc-crawler seems to be very dumb at this and recreates whole > database (haven't seen sources but it takes ages and does a lot of write > i/o after reboot or card reinsert whne there is no change in data on the > card). I think it's supposed just to stat files and read them only if they have changed. If it's not doing this, please file a bug. > Another hint could be to check free available space, if it is same like > before it is unlikely there was some change and even recursive scan can > be skipped. This won't really work. Although it's unlikely, all the content on the MMC could have changed while the amount of used space is same. > So the idea is not to have perfect implementation but fast one solving > most frequent cases (1. no change, 2. just files added/removed) with > possibility to do manual rescan. Is it possible that some specific file(s) are triggering some bug in libid3 that causes it to scan the whole mp3 instead of just the metadata part? If you notice with: for p in $(pidof metalayer-crawler); do ls -l /proc/$p/fd/|grep mmc; done That metalayer-crawler is spending an inordinate amount on some specific files on the MMC, could you try removing that file and checking whether the problem is triggered again? - Eero ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
On Tue, May 22, 2007 at 10:31:49AM -0300, Gustavo Sverzut Barbieri wrote: > > Same applies to metalayer-crawler of course. > > I hope in future we can just use device-wide db and do no scanning at > all. If metalayer-crawler export its data somehow, I'd be really glad > to remove the scanning from Canola, it no benefit having it on our > side. There's an sqlite database in /home/user: $ sudo apt-get install sqlite3 ... Get:1 http://repository.maemo.org bora/free sqlite3 3.3.5-0.2osso1 [21.5kB] ... $ sqlite3 .meta_storage sqlite> .headers on sqlite> select * from Metadata limit 1; Filename|Album|Title|Year|Publisher|Artist|Genre|Track|Bitrate|Length|MIME|Resolution|Fps|Codec|PixelAspectRatio|Channels|Frequency|AccessTime|Present|UPnP|UDN|Filesize|LastModification|AddedTime|isPlaylist|isStream /home/user/MyDocs/.videos/Nokia_N800.avi||Nokia_N800.avi|||101|video/x-msvideo|400x240|25|DIVX1175523035|1|||9477310|1169725653|1174672960|| Marius Gedminas -- MCSE == Marginal Computer Software Enthusiast signature.asc Description: Digital signature ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Gustavo Sverzut Barbieri wrote: > > I don't like the "timer" thing, but yes, it can help and I've already > considered it. > > As for reboot, remember that on reboot, it's not re-parsed (files are > not opened), just re-scanned (stat and compare mtimes). We must do > this stat in order to check for files that changed elsewhere. Not sure how Canola works. Maybe metalayer-crawler does it. We don't have sources, right? For both we can only guess what you guys do in the code. > > keep db in mmc is problematic, since users want a unified view of > every data and this requires us to merge selects from every db, either > by creating a virtual db with data or doing that with code (our own > virtual db). That creates more problems than it saves. Maybe. Don't know implementation. I just know that if some media is not accessible there is no point of keeping the metadata so having them together with data seems to make sense. Even if it is moderately complex it may be worth the effort since it seems like the right thing (to have more metadata databases). This may even allow you to share the index with some desktop application when you use the card with usb reader and even create index on PC to save tablet power (which is how itunes may work). > >> You can also do MD5 sum of directory listing (even output of ls -R or >> -lR) and do not scan files at all if listing hash is same. > > this is easy, but not much faster than the stats... Sure if you stat directories recursively it is same. The idea was just to avoid reading the file data as much as possible. At least metalayerc-crawler seems to be very dumb at this and recreates whole database (haven't seen sources but it takes ages and does a lot of write i/o after reboot or card reinsert whne there is no change in data on the card). Another hint could be to check free available space, if it is same like before it is unlikely there was some change and even recursive scan can be skipped. > and you still may > have inconsistencies, imagine you noticed incorrect id3 and then you > plugged in your device using USB cable (your media get unmounted), use > some app on your pc to fix those incorrect id3 and then remove the > cable, you want your data to be updated. I think you (not just Canola, metalayer-crawler too) are penalizing user for many scenarios that are very unlikely causing the most frequent use case to be slow up to the point of being unusable. In such unlikely situation user should always have option to manually do the rescan from control panel applet (possibly also with selecting subtree with the change). So the idea is not to have perfect implementation but fast one solving most frequent cases (1. no change, 2. just files added/removed) with possibility to do manual rescan. There could be even some slider in control panel applet with fast-slower-accurate choice. > I hope in future we can just use device-wide db and do no scanning at > all. If metalayer-crawler export its data somehow, I'd be really glad > to remove the scanning from Canola, it no benefit having it on our > side. Definitely, you guys should talk together and do one best implementation :-) Frantisek ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
On 5/21/07, Frantisek Dufka <[EMAIL PROTECTED]> wrote: > Gustavo Sverzut Barbieri wrote: > > > > Just hack scripts (both init.d and dbus/service) and launch it with > > lower priority, it should be the default in next release. > > This still puts big stress on memory subsystem and slows device by > flushing caches. :-( > > - data is deleted when SD is removed in order to avoid DB growing > > indefinitely, maybe you'll not see the same card again (it was a > > friend's SD, for example) or data there will not be present anytime > > soon (you just deleted those files using PC or other device) > > You could have option to purge old data when card was not seen for some > time (like few days). It is much better then scanning card again when > rebooting or quick removal and reinsert (can be quite frequent with some > use cases). I don't like the "timer" thing, but yes, it can help and I've already considered it. As for reboot, remember that on reboot, it's not re-parsed (files are not opened), just re-scanned (stat and compare mtimes). We must do this stat in order to check for files that changed elsewhere. > Also if you fear of huge DB size you may consider to optionally keep DB > directly on card with media files if it is writable and there is free > space and also user sets it in preferences (or even if user moves it by > hand there if canola is smart enough to keep one DB per scanned card, > why not?). keep db in mmc is problematic, since users want a unified view of every data and this requires us to merge selects from every db, either by creating a virtual db with data or doing that with code (our own virtual db). That creates more problems than it saves. > You can also do MD5 sum of directory listing (even output of ls -R or > -lR) and do not scan files at all if listing hash is same. this is easy, but not much faster than the stats... and you still may have inconsistencies, imagine you noticed incorrect id3 and then you plugged in your device using USB cable (your media get unmounted), use some app on your pc to fix those incorrect id3 and then remove the cable, you want your data to be updated. > Same applies to metalayer-crawler of course. I hope in future we can just use device-wide db and do no scanning at all. If metalayer-crawler export its data somehow, I'd be really glad to remove the scanning from Canola, it no benefit having it on our side. -- Gustavo Sverzut Barbieri -- Jabber: [EMAIL PROTECTED] MSN: [EMAIL PROTECTED] ICQ#: 17249123 Skype: gsbarbieri Mobile: +55 (81) 9927 0010 ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
ext Eero Tamminen wrote: > Hi, > > ext Gustavo Sverzut Barbieri wrote: > >>> I use two 4gig SD cards. Once i stuffed them full of media canola-conf >>> would just take over my cpu ( i am taking a stab in the dark, that >>> this was the indexer) . I waited for quite a while, and even had to >>> charge my n800 again, and i still couldn't use it, as everything would >>> stop responding. >>> >>> I just uninstalled it as i got sick of doing 'killall -9 canola-conf' >>> via xterm everytime my device slowed right down. >>> >>> If there is anything i can do to help debug this, then let me know and >>> i will try again, but I am also using a modified kernel to support >>> SDHC cards, so I don't know if my debugging counts :) >>> >> Just hack scripts (both init.d and dbus/service) and launch it with >> lower priority, it should be the default in next release. >> >> It basically works like this: >> - 1st time: readdir() + regexp (defined in gconf) + open() + parse >> (id3lib, shipped with device) + db insert >> > > I have strong suspicion that the culprit is libid3. When stracing on > the device what happens when MMC is scanned, the files are read 8KB > at the time, then the file offset is moved one byte and again 8KB are > read etc. > > I think it would be nicer for such scanners to take just the filepath and do the libid3 parsing later on, user won't see those information anyway before the UI is shown. This kind of 'lazy setup' would ease the amount of work needed for scanner. UI should then take care not to block too much on libid3 operations but try to do it in another thread while viewing playlist. > If you can test Canola on x86, you could try ltracing what happens > (ltrace provides more info as it traces also exported library functions, > but it doesn't work on ARM). > > > - Eero > ___ > maemo-developers mailing list > maemo-developers@maemo.org > https://maemo.org/mailman/listinfo/maemo-developers > > // Tapani ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Hi, ext Gustavo Sverzut Barbieri wrote: >> I use two 4gig SD cards. Once i stuffed them full of media canola-conf >> would just take over my cpu ( i am taking a stab in the dark, that >> this was the indexer) . I waited for quite a while, and even had to >> charge my n800 again, and i still couldn't use it, as everything would >> stop responding. >> >> I just uninstalled it as i got sick of doing 'killall -9 canola-conf' >> via xterm everytime my device slowed right down. >> >> If there is anything i can do to help debug this, then let me know and >> i will try again, but I am also using a modified kernel to support >> SDHC cards, so I don't know if my debugging counts :) > > Just hack scripts (both init.d and dbus/service) and launch it with > lower priority, it should be the default in next release. > > It basically works like this: > - 1st time: readdir() + regexp (defined in gconf) + open() + parse > (id3lib, shipped with device) + db insert I have strong suspicion that the culprit is libid3. When stracing on the device what happens when MMC is scanned, the files are read 8KB at the time, then the file offset is moved one byte and again 8KB are read etc. If you can test Canola on x86, you could try ltracing what happens (ltrace provides more info as it traces also exported library functions, but it doesn't work on ARM). - Eero ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Awesome! Thanks for the tips, i will have another go :) Cheers, and a great app by the way! On 5/21/07, Gustavo Sverzut Barbieri <[EMAIL PROTECTED]> wrote: > Damn, I forgot to give you more useful hints on some specific issues! > > On 5/21/07, nick loeve <[EMAIL PROTECTED]> wrote: > > I use two 4gig SD cards. Once i stuffed them full of media canola-conf > > would just take over my cpu ( i am taking a stab in the dark, that > > this was the indexer) . I waited for quite a while, and even had to > > charge my n800 again, and i still couldn't use it, as everything would > > stop responding. > > > > I just uninstalled it as i got sick of doing 'killall -9 canola-conf' > > via xterm everytime my device slowed right down. > > Just use canola applet, it will give you hints when canola-conf is > scanning collection and you can click it in order to abort this > operation... also enabling you to trigger it later. > > This is all using dbus signals, so maybe you can issue dbus-send > commands by hand or some other script, app if you wish. > > -- > Gustavo Sverzut Barbieri > -- > Jabber: [EMAIL PROTECTED] >MSN: [EMAIL PROTECTED] > ICQ#: 17249123 > Skype: gsbarbieri > Mobile: +55 (81) 9927 0010 > -- Nick Loeve www.trickie.org ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Gustavo Sverzut Barbieri wrote: > > Just hack scripts (both init.d and dbus/service) and launch it with > lower priority, it should be the default in next release. This still puts big stress on memory subsystem and slows device by flushing caches. > - data is deleted when SD is removed in order to avoid DB growing > indefinitely, maybe you'll not see the same card again (it was a > friend's SD, for example) or data there will not be present anytime > soon (you just deleted those files using PC or other device) You could have option to purge old data when card was not seen for some time (like few days). It is much better then scanning card again when rebooting or quick removal and reinsert (can be quite frequent with some use cases). Also if you fear of huge DB size you may consider to optionally keep DB directly on card with media files if it is writable and there is free space and also user sets it in preferences (or even if user moves it by hand there if canola is smart enough to keep one DB per scanned card, why not?). You can also do MD5 sum of directory listing (even output of ls -R or -lR) and do not scan files at all if listing hash is same. Same applies to metalayer-crawler of course. > > it's always a trade off Well, sometimes it can be also sloppy coding ;-) Frantisek ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Damn, I forgot to give you more useful hints on some specific issues! On 5/21/07, nick loeve <[EMAIL PROTECTED]> wrote: > I use two 4gig SD cards. Once i stuffed them full of media canola-conf > would just take over my cpu ( i am taking a stab in the dark, that > this was the indexer) . I waited for quite a while, and even had to > charge my n800 again, and i still couldn't use it, as everything would > stop responding. > > I just uninstalled it as i got sick of doing 'killall -9 canola-conf' > via xterm everytime my device slowed right down. Just use canola applet, it will give you hints when canola-conf is scanning collection and you can click it in order to abort this operation... also enabling you to trigger it later. This is all using dbus signals, so maybe you can issue dbus-send commands by hand or some other script, app if you wish. -- Gustavo Sverzut Barbieri -- Jabber: [EMAIL PROTECTED] MSN: [EMAIL PROTECTED] ICQ#: 17249123 Skype: gsbarbieri Mobile: +55 (81) 9927 0010 ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
On 5/21/07, nick loeve <[EMAIL PROTECTED]> wrote: > I use two 4gig SD cards. Once i stuffed them full of media canola-conf > would just take over my cpu ( i am taking a stab in the dark, that > this was the indexer) . I waited for quite a while, and even had to > charge my n800 again, and i still couldn't use it, as everything would > stop responding. > > I just uninstalled it as i got sick of doing 'killall -9 canola-conf' > via xterm everytime my device slowed right down. > > If there is anything i can do to help debug this, then let me know and > i will try again, but I am also using a modified kernel to support > SDHC cards, so I don't know if my debugging counts :) Just hack scripts (both init.d and dbus/service) and launch it with lower priority, it should be the default in next release. It basically works like this: - 1st time: readdir() + regexp (defined in gconf) + open() + parse (id3lib, shipped with device) + db insert - on reboot: consistency check: db select + stat(), readdir() + regexp + db query + parse (if not in db) - on SD removal: db delete from table where path like '/mnt/mmcX/%'. - on SD add: consider as first time. Points to consider: - collection can be restricted, just use configuration tool. Future versions will allow you to limit parsers (media type) per folder, like /mnt/mmcX/Audio is just for audio, leave my covers out of picture collection. - data is deleted when SD is removed in order to avoid DB growing indefinitely, maybe you'll not see the same card again (it was a friend's SD, for example) or data there will not be present anytime soon (you just deleted those files using PC or other device) it's always a trade off, you can also open xterm and execute gst-launch... or use any simple UI. -- Gustavo Sverzut Barbieri -- Jabber: [EMAIL PROTECTED] MSN: [EMAIL PROTECTED] ICQ#: 17249123 Skype: gsbarbieri Mobile: +55 (81) 9927 0010 ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
I use two 4gig SD cards. Once i stuffed them full of media canola-conf would just take over my cpu ( i am taking a stab in the dark, that this was the indexer) . I waited for quite a while, and even had to charge my n800 again, and i still couldn't use it, as everything would stop responding. I just uninstalled it as i got sick of doing 'killall -9 canola-conf' via xterm everytime my device slowed right down. If there is anything i can do to help debug this, then let me know and i will try again, but I am also using a modified kernel to support SDHC cards, so I don't know if my debugging counts :) Cheers On 5/21/07, Felipe Contreras <[EMAIL PROTECTED]> wrote: > On 5/21/07, Karoliina Salminen <[EMAIL PROTECTED]> wrote: > > Hi, > > > > > > ext Kalle Vahlman wrote: > > > > > >> Folder approach is intuitive, shared by all reasonable apps on all > > >> platforms, doesn't waste anything and just works. > > >> > > > > > > It does waste the users time (in the long run), which to me seems more > > > important than the machines time. > > > > > It pretty much depends on the user and if there are mp3-tags present on > > the files, I would like to have > > both ways available because: > > I happen to have a tag-messy music library (which is organized to > > folders, which is not my fault - I have done a lots of work to make > > compilations of various mp3-files I have got from my artist friends, but > > the tags being inconsistent is the fault of > > the author and encoder of the mp3-files) because my library consists of > > mostly indie music done by > > some friends. Obviously there are no mp3-tags on place and the files are > > organized with > > folders only (as the author of the file haven't been too interested to > > put the tags on place especially > > if the file is unfinished track that is just given out for > > review-listening), playing full albums (which are > > based on these files without mp3-tags or varying mp3-tags like Track 1 > > "Fin:Greenblue 1" > > Track 2 "fin: Green blue 2" Track 3 "FIN: Blue Green3" Track 4 > > "Christian Worton:greenblue 4" Track 5 "Fin: Greenblue 5" Track 6: > > "Unknown artist:Green Blue 6" ... > > and so forth. This ends up being several artists and several albums on > > system which blindly indexes the files based on the mp3 tags, which is > > pretty problematic and which isn't my fault as listener and > > it makes me pretty confused. The only solution currently to my > > understanding, in media players only supporting > > mp3-tag based organization is to play the files one by one which is not > > so nice for listening to > > background music. > > > > Now I could imagine a brilliant tool that would sort of solve this problem: > > A program or script that would run on linux and which would go and scan > > all the subfolders in the media > > folder and put the tags to the mp3-files on place so that the album name > > would be created from the folder name and the track name from the > > filename. Obviously it should do this without re-encoding the files > > (since re-encoding would degrade sound quality in lossy compression). If > > such tool already exists, I would love to hear where I can get it, I > > would really > > need one since I can't really blame the various authors of my > > mp3-collection since if I am getting better indie music than > > commercial music is for free, I can't really complain, having > > inconsistent mp3-tags is small issue taken in account that > > the music is superb quality, glitches on tags, but no glitches on music! > > There are some tools: > > http://pwp.netcabo.pt/paol/tagtool/ > http://easytag.sourceforge.net/ > http://www.gnomefiles.org/app.php/gmusicbrowser > > Best regards. > > -- > Felipe Contreras > ___ > maemo-developers mailing list > maemo-developers@maemo.org > https://maemo.org/mailman/listinfo/maemo-developers > -- Nick Loeve www.trickie.org ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
On 5/21/07, Karoliina Salminen <[EMAIL PROTECTED]> wrote: > Hi, > > > ext Kalle Vahlman wrote: > > > >> Folder approach is intuitive, shared by all reasonable apps on all > >> platforms, doesn't waste anything and just works. > >> > > > > It does waste the users time (in the long run), which to me seems more > > important than the machines time. > > > It pretty much depends on the user and if there are mp3-tags present on > the files, I would like to have > both ways available because: > I happen to have a tag-messy music library (which is organized to > folders, which is not my fault - I have done a lots of work to make > compilations of various mp3-files I have got from my artist friends, but > the tags being inconsistent is the fault of > the author and encoder of the mp3-files) because my library consists of > mostly indie music done by > some friends. Obviously there are no mp3-tags on place and the files are > organized with > folders only (as the author of the file haven't been too interested to > put the tags on place especially > if the file is unfinished track that is just given out for > review-listening), playing full albums (which are > based on these files without mp3-tags or varying mp3-tags like Track 1 > "Fin:Greenblue 1" > Track 2 "fin: Green blue 2" Track 3 "FIN: Blue Green3" Track 4 > "Christian Worton:greenblue 4" Track 5 "Fin: Greenblue 5" Track 6: > "Unknown artist:Green Blue 6" ... > and so forth. This ends up being several artists and several albums on > system which blindly indexes the files based on the mp3 tags, which is > pretty problematic and which isn't my fault as listener and > it makes me pretty confused. The only solution currently to my > understanding, in media players only supporting > mp3-tag based organization is to play the files one by one which is not > so nice for listening to > background music. > > Now I could imagine a brilliant tool that would sort of solve this problem: > A program or script that would run on linux and which would go and scan > all the subfolders in the media > folder and put the tags to the mp3-files on place so that the album name > would be created from the folder name and the track name from the > filename. Obviously it should do this without re-encoding the files > (since re-encoding would degrade sound quality in lossy compression). If > such tool already exists, I would love to hear where I can get it, I > would really > need one since I can't really blame the various authors of my > mp3-collection since if I am getting better indie music than > commercial music is for free, I can't really complain, having > inconsistent mp3-tags is small issue taken in account that > the music is superb quality, glitches on tags, but no glitches on music! There are some tools: http://pwp.netcabo.pt/paol/tagtool/ http://easytag.sourceforge.net/ http://www.gnomefiles.org/app.php/gmusicbrowser Best regards. -- Felipe Contreras ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Am Montag 21 Mai 2007 00:13 schrieb Karoliina Salminen: > > Now I could imagine a brilliant tool that would sort of solve this problem: > A program or script that would run on linux and which would go and scan > all the subfolders in the media > folder and put the tags to the mp3-files on place so that the album name > would be created from the folder name and the track name from the > filename. Obviously it should do this without re-encoding the files > (since re-encoding would degrade sound quality in lossy compression). If > such tool already exists, I would love to hear where I can get it There are several commandline tools that do things like that, what about this one (you can find more with "apt-cache search mp3" on a Debian system): $ apt-cache show mp3info Package: mp3info Priority: extra Section: sound Installed-Size: 76 Maintainer: Pawel Wiecek <[EMAIL PROTECTED]> Architecture: i386 Version: 0.8.4-9.2 Depends: libc6 (>= 2.3.6-6), libncurses5 (>= 5.4-5), debconf | debconf-2.0 Filename: pool/main/m/mp3info/mp3info_0.8.4-9.2_i386.deb Size: 30294 MD5sum: 1ba2865b321af42f089bb823b500671c SHA1: 1f459ae83736e9d00fb2af8f11061e7e50fe3053 SHA256: 38de501c7872d7c0078a6158da0ecd52d978da5badd36785cb6e4a906813e7d9 Description: An MP3 technical info viewer and ID3 1.x tag editor MP3Info has an interactive mode (using curses) and a command line mode. MP3Info can display ID3 tag information as well as various technical aspects of an MP3 file including playing time, bit-rate, sampling frequency and other attributes in a pre-defined or user-specifiable output format. . If you prefer GUI you should use mp3info-gtk package. Tag: interface::text-mode, role::program, scope::utility, uitoolkit::ncurses, use::checking, works-with::audio, works-with-format::mp3 HTH, Hans ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Hi, ext Kalle Vahlman wrote: > >> Folder approach is intuitive, shared by all reasonable apps on all >> platforms, doesn't waste anything and just works. >> > > It does waste the users time (in the long run), which to me seems more > important than the machines time. > It pretty much depends on the user and if there are mp3-tags present on the files, I would like to have both ways available because: I happen to have a tag-messy music library (which is organized to folders, which is not my fault - I have done a lots of work to make compilations of various mp3-files I have got from my artist friends, but the tags being inconsistent is the fault of the author and encoder of the mp3-files) because my library consists of mostly indie music done by some friends. Obviously there are no mp3-tags on place and the files are organized with folders only (as the author of the file haven't been too interested to put the tags on place especially if the file is unfinished track that is just given out for review-listening), playing full albums (which are based on these files without mp3-tags or varying mp3-tags like Track 1 "Fin:Greenblue 1" Track 2 "fin: Green blue 2" Track 3 "FIN: Blue Green3" Track 4 "Christian Worton:greenblue 4" Track 5 "Fin: Greenblue 5" Track 6: "Unknown artist:Green Blue 6" ... and so forth. This ends up being several artists and several albums on system which blindly indexes the files based on the mp3 tags, which is pretty problematic and which isn't my fault as listener and it makes me pretty confused. The only solution currently to my understanding, in media players only supporting mp3-tag based organization is to play the files one by one which is not so nice for listening to background music. Now I could imagine a brilliant tool that would sort of solve this problem: A program or script that would run on linux and which would go and scan all the subfolders in the media folder and put the tags to the mp3-files on place so that the album name would be created from the folder name and the track name from the filename. Obviously it should do this without re-encoding the files (since re-encoding would degrade sound quality in lossy compression). If such tool already exists, I would love to hear where I can get it, I would really need one since I can't really blame the various authors of my mp3-collection since if I am getting better indie music than commercial music is for free, I can't really complain, having inconsistent mp3-tags is small issue taken in account that the music is superb quality, glitches on tags, but no glitches on music! Best Regards, Karoliina ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
On Sun, May 20, 2007 at 09:03:31PM +0200, ext Laurent GUERBY wrote: > On Sun, 2007-05-20 at 20:45 +0300, Daniel Stone wrote: > > On Sun, May 20, 2007 at 04:25:04PM +0200, ext Laurent GUERBY wrote: > > > Folder approach is intuitive, shared by all reasonable apps on all > > > platforms > > > > Except for more or less every media player ever made (cf. iTunes). > > Amarok is frequently cited as a free software iTunes equivalent and it > works *exactly* as I describe: if you want it to index you can choose > what folder(s) to index Canola also works this way. > It's quite easy to find people complaining about iTunes being > a ressource hog and taking forever to scan stuff. Sure, but stop any random person on the street, and ask them about the iTunes UI. See if you have the same reaction. > And I still can't honestly believe that the developpers of media > software want to tell their user: oh you have a 2GB card fully > of media and you want to play it on your N800? No problem! Insert > it then please wait two hours ... and no sorry you can't use your > N800 because it's dog slow ... then there you go! How great and easy! I use an N800 with two 4GB cards, both containing only media (and full of same), with Canola. Works pretty okay for me. signature.asc Description: Digital signature ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
On Sun, 2007-05-20 at 22:31 +0300, Kalle Vahlman wrote: > 2007/5/20, Laurent GUERBY <[EMAIL PROTECTED]>: > > Today I installed canola. Removed it after 10 minutes. > > > > Why are developpers of media application using the "scan+database" based > > approach instead of just letting user browse folders to open > > the media they want? > > Because people are lazy and want the machine do the organizing for > them. That's why we have metadata in the files, so that the user can > safely keep the files in gigantic blobs and still find the correct one > when they want to, without scanning the thousands of crappily named > files over. Note that I don't say the indexing isn't useful "If you want indexing for the user just add a "add media directory to index" option and let user open "index" files." I'm saying proposing *only* indexing (and total from / indexing...) is completely wrong. Indexing *is* of course useful (especially when meta data is present), I fully agree with you on that. But so is being to use the software on a mobility device without waiting two hours :). Laurent ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Am Sonntag 20 Mai 2007 16:25 schrieb Laurent GUERBY: > Today I installed canola. Removed it after 10 minutes. > > Why are developpers of media application using the "scan+database" based > approach instead of just letting user browse folders to open > the media they want? I fully agree. Thanks for your hint, you saved me some time. I find this database approach questionable on a desktop PC, and it's certainly complete crap on a device like the 770/800. Hans ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
2007/5/20, Laurent GUERBY <[EMAIL PROTECTED]>: > Today I installed canola. Removed it after 10 minutes. > > Why are developpers of media application using the "scan+database" based > approach instead of just letting user browse folders to open > the media they want? Because people are lazy and want the machine do the organizing for them. That's why we have metadata in the files, so that the user can safely keep the files in gigantic blobs and still find the correct one when they want to, without scanning the thousands of crappily named files over. So in short, because people want them to do it. That said, for an power-sensitive situation, the indexing should be very conservative (which I guess the current ones are not?). > The scan+database approach: > > - wastes precious energy on mobile devices (99% of the scan will be > wasted and I assume reading the card is not free) As I said above. > - slow things down by wasting system cache (useless IO) I guess... > - doesn't work at all when the user has multiple cards holding well ... > media (why would one buy multiple cards otherwise?) Why not? The cards have an unique ID to which the player could link the media on it, and hide the media that's on a non-present card. Surely this is considered in any indexing system, right? ;) > - doesn't work when the user reorganize its folders (start scan again) I think "doesn't work" is a bit strongly said here, it's just extra work... > - prevents user to remove cards "in use" with scary messages That's bad. But I haven't seen this though, is it a common problem? > - doesn't work when the card has many non media folders and files (eg: > maemo mapper which is a great application) Again, IMO "doesn't work" is a bit off. > - presents the user with an awful install experience: "please wait" and > you can't do a thing for minutes (or hours I can't say). > - is totally user-interface incoherent with about all the other apps on > the device > > Folder approach is intuitive, shared by all reasonable apps on all > platforms, doesn't waste anything and just works. It does waste the users time (in the long run), which to me seems more important than the machines time. > How long before common sense returns? I bet there are players that can be used without the media library approach (like the already mentioned Kilikali), but I doubt that will be a winning trend. In fact, I expect the folder approach (as it has been since the eighties) to give away for more powerful (and yes, more power consuming at the same time) approaches like searching/indexing and contextual linking (as opposed to navigational linking). That's how the massive amounts of data on the web is handled, and that's how it is going to be on the desktop too. And, since you are not organized on the desktop, you won't be on your handheld either so But those are only my speculations naturally. -- Kalle Vahlman, [EMAIL PROTECTED] Powered by http://movial.fi Interesting stuff at http://syslog.movial.fi ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
On Sun, 2007-05-20 at 20:45 +0300, Daniel Stone wrote: > On Sun, May 20, 2007 at 04:25:04PM +0200, ext Laurent GUERBY wrote: > > Folder approach is intuitive, shared by all reasonable apps on all > > platforms > > Except for more or less every media player ever made (cf. iTunes). Amarok is frequently cited as a free software iTunes equivalent and it works *exactly* as I describe: if you want it to index you can choose what folder(s) to index, if you don't want to index you just use "open" and it works just like any UI standard conforming software. Intuitive, reactive and powerful. It's quite easy to find people complaining about iTunes being a ressource hog and taking forever to scan stuff. However biased it is, google "itunes amarok". And I still can't honestly believe that the developpers of media software want to tell their user: oh you have a 2GB card fully of media and you want to play it on your N800? No problem! Insert it then please wait two hours ... and no sorry you can't use your N800 because it's dog slow ... then there you go! How great and easy! Spot the problem. Laurent ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
Laurent GUERBY wrote: > Today I installed canola. Removed it after 10 minutes. > > Why are developpers of media application using the "scan+database" based > approach Try kilikali music player, if you want something simple but working: http://kilikali.garage.maemo.org/ -- Tuomas signature.asc Description: OpenPGP digital signature ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
Re: What's wrong with folder browsing?
On Sun, May 20, 2007 at 04:25:04PM +0200, ext Laurent GUERBY wrote: > Folder approach is intuitive, shared by all reasonable apps on all > platforms Except for more or less every media player ever made (cf. iTunes). signature.asc Description: Digital signature ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
What's wrong with folder browsing?
Today I installed canola. Removed it after 10 minutes. Why are developpers of media application using the "scan+database" based approach instead of just letting user browse folders to open the media they want? The scan+database approach: - wastes precious energy on mobile devices (99% of the scan will be wasted and I assume reading the card is not free) - slow things down by wasting system cache (useless IO) - doesn't work at all when the user has multiple cards holding well ... media (why would one buy multiple cards otherwise?) - doesn't work when the user reorganize its folders (start scan again) - prevents user to remove cards "in use" with scary messages - doesn't work when the card has many non media folders and files (eg: maemo mapper which is a great application) - presents the user with an awful install experience: "please wait" and you can't do a thing for minutes (or hours I can't say). - is totally user-interface incoherent with about all the other apps on the device Folder approach is intuitive, shared by all reasonable apps on all platforms, doesn't waste anything and just works. If you want indexing for the user just add a "add media directory to index" option and let user open "index" files. Any idea? How long before common sense returns? Sorry for the rant, Laurent ___ maemo-developers mailing list maemo-developers@maemo.org https://maemo.org/mailman/listinfo/maemo-developers
