All Projects

$ make bdf2bmp
cc -O -g -W -Wall -Wshadow -pedantic -g bdf2bmp.c -o bdf2bmp
bdf2bmp.c:100:5: error: conflicting types for ‘getline’
In file included from bdf2bmp.c:49:0:
/usr/include/stdio.h:671:20: note: previous declaration of ‘getline’ was here
bdf2bmp.c:457:5: error: conflicting types for ‘getline’
In file included from bdf2bmp.c:49:0:
/usr/include/stdio.h:671:20: note: previous declaration of ‘getline’ was here

Renaming the function getline fixes this.

Updated Danish language file

This patch adds a menu entry to the debug menu which will allow the user to toggle writing of a log of files being read for metadata in /metadata.log. The filename is written before Rockbox attempts to read metadata from the file, so if Rockbox crashes when scanning for the DB, the last filename in the log should be the file that caused the crash. This should greatly help debug metadata reader problems.

This patch makes the theme site check if an uploaded theme contains a font which is already in the font pack, and rejects it if it does.

This simple patch stops set_file from modifying the filename argument.

Line 1157 sets the last / to \0, to split the filename variable into dir and basename. However, the dir part is only used in the following statement: strncasecmp(ROCKBOX_DIR, filename ,strlen(ROCKBOX_DIR)

Here, the comparison will be either true or false even without splitting the filename variable into dir part and basename part. I suspect it was previously doing a “dumber” comparison, and someone changed it to use strncasecmp, and didn’t see this issue.

This is needed to fix a bug in  FS#10931  that would cause themes to not load at all.

This patch will center on the most recently loaded theme when browsing themes. If the user has changed theme related settings, this will not be completely accurate. In my opinion, this will very very rarely lead to confusion, since the user will know that he’s changed some settings. It could even be seen as a quick way to “reset” theme settings to the one he started out with.

It also changes “Browse Theme Files” to simply “Theme”, to make it feel more like a setting.

Building checkwps (up to date SVN r21295) fails for all targets with errors such as this:

CC [IRIVER_H120]
In file included from ../../apps/misc.h:27,

               from ../../apps/gui/wps_parser.c:26:

../../firmware/export/system.h:151:1: warning: “htole16” redefined
In file included from /usr/include/sys/types.h:217,

               from /usr/include/stdlib.h:320,
               from ../../apps/gui/wps_parser.c:24:

/usr/include/endian.h:65:1: warning: this is the location of the previous definition
In file included from ../../apps/misc.h:27,

               from ../../apps/gui/wps_parser.c:26:

../../firmware/export/system.h:152:1: warning: “htole32” redefined
In file included from /usr/include/sys/types.h:217,

               from /usr/include/stdlib.h:320,
               from ../../apps/gui/wps_parser.c:24:

/usr/include/endian.h:70:1: warning: this is the location of the previous definition
In file included from ../../apps/misc.h:27,

               from ../../apps/gui/wps_parser.c:26:

../../firmware/export/system.h:155:1: warning: “htobe16” redefined
In file included from /usr/include/sys/types.h:217,

               from /usr/include/stdlib.h:320,
               from ../../apps/gui/wps_parser.c:24:

/usr/include/endian.h:64:1: warning: this is the location of the previous definition
In file included from ../../apps/misc.h:27,

               from ../../apps/gui/wps_parser.c:26:

../../firmware/export/system.h:156:1: warning: “htobe32” redefined
In file included from /usr/include/sys/types.h:217,

               from /usr/include/stdlib.h:320,
               from ../../apps/gui/wps_parser.c:24:

/usr/include/endian.h:69:1: warning: this is the location of the previous definition
In file included from ../../apps/misc.h:27,

               from ../../apps/gui/wps_parser.c:26:

../../firmware/export/system.h:232:27: error: system-target.h: No such file or directory
In file included from ../../apps/gui/wps_parser.c:34:
./checkwps.h:42:1: warning: “MIN” redefined
In file included from ../../apps/misc.h:27,

               from ../../apps/gui/wps_parser.c:26:

../../firmware/export/system.h:90:1: warning: this is the location of the previous definition
In file included from ../../apps/gui/wps_parser.c:47:
../../apps/settings.h:290: error: redefinition of ‘struct user_settings’ ../../apps/settings.h:743: error: conflicting types for ‘global_settings’ ./checkwps.h:36: error: previous declaration of ‘global_settings’ was here
In file included from ../../apps/misc.h:27,

               from ../../apps/misc.c:25:

../../firmware/export/system.h:151:1: warning: “htole16” redefined
In file included from /usr/include/sys/types.h:217,

               from /usr/include/stdlib.h:320,
               from ../../apps/misc.c:21:

/usr/include/endian.h:65:1: warning: this is the location of the previous definition
In file included from ../../apps/misc.h:27,

               from ../../apps/misc.c:25:

../../firmware/export/system.h:152:1: warning: “htole32” redefined
In file included from /usr/include/sys/types.h:217,

               from /usr/include/stdlib.h:320,
               from ../../apps/misc.c:21:

/usr/include/endian.h:70:1: warning: this is the location of the previous definition
In file included from ../../apps/misc.h:27,

               from ../../apps/misc.c:25:

../../firmware/export/system.h:155:1: warning: “htobe16” redefined
In file included from /usr/include/sys/types.h:217,

               from /usr/include/stdlib.h:320,
               from ../../apps/misc.c:21:

/usr/include/endian.h:64:1: warning: this is the location of the previous definition
In file included from ../../apps/misc.h:27,

               from ../../apps/misc.c:25:

../../firmware/export/system.h:156:1: warning: “htobe32” redefined
In file included from /usr/include/sys/types.h:217,

               from /usr/include/stdlib.h:320,
               from ../../apps/misc.c:21:

/usr/include/endian.h:69:1: warning: this is the location of the previous definition
In file included from ../../apps/misc.h:27,

               from ../../apps/misc.c:25:

../../firmware/export/system.h:232:27: error: system-target.h: No such file or directory
make: *** [checkwps] Error 1

This is rather unfortunate for the theme site, which uses checkwps to check if themes are valid.

This patch centers the browser on the currently loaded font when doing “browse fonts”.

Something smarter should probably be done that does the same for themes, languages and wps files.

Rockbox doesn’t read the tags of this file.

It contains the following tags:

ID3v1 tag

Title: Now Get Busy
Artist: Beastie Boys
Album: The Wired CD
Year: 2004
Track: 1
Genre: Hip-Hop

ID3v2.4.0 tag

replaygain_album_gain (TXXX): -7.62 dB
replaygain_album_peak (TXXX): 1.278176665
replaygain_track_peak (TXXX): 1.033416986
replaygain_track_gain (TXXX): -8.38 dB

However, Rockbox only reads the v2 tag, never bothering to check for the artist, title, album etc tag in the v1 tag.

Track is licensed CC-NC-Sampling+ 1.0 and copyright Beastie Boys, obviously.

This patch should allow the user to select - at configure-time - which of his installed festival voices to use. I don’t have more than one voice installed at the moment, so I haven’t done any meaningful test yet.

Please test and report back.

These are bootloaders for the PP targets which will be released with USB enabled for 3.2. As the main feature, they do not boot the OF when booting with USB inserted.

Please test and report back. The zip contain all bootloaders placed in a directory structure that should be self-explanatory.

Tested BROKEN:

Tested OK:
- Ipod 4G
- Ipod Video
- Iriver H10-5GB
- M:Robe 100
- Sansa C200
- Sansa E200
- Sansa E200R
- Iriver H10 (20GB)

Needs testing:

It turns out that these bootloaders haven’t changed behavior for Ipods, so no need to test those, as we won’t be releasing updated bootloaders for Ipods.

When copying a file from one volume to another, the disk full condition isn’t detected, and the copying never finishes.

I used the following method:
* Create 2 files on the player - one 110MB file of zeroes (0×0) and a 20MB file of ones (0×1)
* Copy the 110MB file to a 128MB microsd card
* Copy the 20MB file

The latter operation never finished (after 15 minutes), and I had to forcefully shut down by holding power for 15 seconds. After mounting the card, I see the 110MB file transferred correctly (MD5 sum matched), and a 0-byte file with the filename of the 20MB file.

Users on http://anythingbutipod.com/forum/showthread.php?t=34458 hypothesized that corruption might occur, but this doesn’t seem to be the case. At least for this 128MB microsd card.

Tested on a Sansa E280 - I expect this might be true for all multivolume targets.

This patch adds a section describing the viewport tags to the manual.

There are a few fixmes, which I’d be happy if someone could look at.

Two relate to using \opt (or \optv) with the example environment, and another is a weird issue that I can’t figure out which means I ended up adding a blank line to a tagmap table.

This patch adds the beginning of an Album Art appendix to the manual. I think an appendix is most appropriate, since not all targets that support Album Art will ever support it in the WPS, so putting it in that chapter would be wrong. A reference would probably be smart though.

A couple of todos/issues with this patch:
* I don’t know how to style the item list/filenames
* The replacement characters get a bit weird. I suppose some escaping is needed, but my LaTeX skills aren’t great.
* It should probably direct users at the Wiki for information on programs to collect/convert albumart.
* I’m not entirely sure which conditions to use in the opt/nopts

The summary pretty much says it all: My usual listening pattern is to add most or all songs on my player, and skip songs I’m not in the mood for. This occasionally results in a lot of skipping (past say, 5-10 tracks), and while I do this the next track info will stay the same, so if I have a playlist of Track1-Track20, and am listening to Track1, and start skipping, the next track will continue to be Track2. I *think* that the next track info is only updated once I stop skipping and a complete rebuffering to the new location has been done, though this is just speculation.

I’m not 100% sure this happens everytime, but I see it often with my usage pattern.

As the summary says, when the tag doesn’t set a tracknumber, the scrobbler.log contains a weird value instead of simply leaving it out as the format specifies (http://www.audioscrobbler.net/wiki/Portable_Player_Logging): “# If the data for optional fields is not available to you, leave the field blank (\t\t).”

This patch inverses album art in the WPS when LCD Mode is set to inverse. This makes the album art come out as it originally is, which is probably reasonable to do.

The patch also no longer sets the drawmode when drawing albumart, since the drawmode doesn’t affect non-mono bitmaps.

This enables the user to use the “LCD Mode” setting to change a light theme into a dark one, and vice versa, without it ruining album art in the WPS.

Currently only tested in e200 sim.

Comments in WPS files act very strangely if they’re not on a line of their own. The following (and attached) wps demonstrates this:

———-Start———-
Artist: %ia
Title: %it
foo
bar
Artist: %ia# Comment
Title: %it# Comment
———–End———–

Expected output:
———-Start———-
Artist: Dire Straits
Title: Walk Of Life
foo
bar
Artist: Dire Straits
Title: Walk Of Life
———–End———–

Actual output:
———-Start———-
Artist: Dire Straits
Title: Walk Of Life
foo
barArtist: Dire StraitsTitle: Walk Of Life
———–End———–

It seems to me that comment-characters should only “eat” line-endings if they’re at the beginning of a line?

This behaviour was introduced in r13140 (http://svn.rockbox.org/viewvc.cgi?view=rev&revision=13140), seemingly on purpose, but I don’t think “inline” comments were considered - that’s why I’m adding this as a bug.

As a note, at least one WPS in SVN uses this behaviour to break a long line, so if it’s fixed, that WPS should be modified as well.

This is a double-size version of 10-Artwiz-Snap, created using bdfresize and manually editing to make it prettier.

It doesn’t have terribly impressive font coverage - same as the currently included snap font. I don’t know if this is wanted.

Edit: Oops, the space character wasn’t empty. Updated.

This patch reverses the logic of the simulator background image, showing it by default and having a –nobackground switch. The rationale is that new users almost invariably have difficulty figuring out how to control the simulator when all they see is the display, whereas experienced users will know to turn it off if it really bothers them.

I’m putting this patch here mostly to get some discussion about whether to do this or not.

Running Pictureflow, it does “Preparing album artwork” for a while, then exits displays two splashes: “No albums found” and “Could not create album art cache”, I assume the latter is because of the former.

I do have plenty of albums and albumart on both internal disk and microsd card. All my albumart is in /.rockbox/albumart/ (available in several sizes, one of which is 100×100).

I suspect this is because id3.artist isn’t set before calling search_albumart_files() in get_albumart_for_index_from_db(), so the search for albums in /.rockbox/albumart/ never runs since it needs to now the artist name. Ideally id3.albumartist should be set as well.

As a sidenote - shouldn’t Pictureflow run even when no albumart is found?

Using r19324.

The attached patch converts all fonts to ISO10646-1 encoding.

This is done using the attached script, or manually for the current ISO8859-1 fonts which simply needed editing of the header, as well as adding appropriate properties to some current ISO10646-1 fonts without any information saying so.

Furthermore, I suggest renaming fonts to make things less confusing:

courB08.bdf ⇒ Courier-Bold-8.bdf
snap.bdf ⇒ snap-10.bdf
win_crox0c.bdf ⇒ Cronyx-Fixed-10.bdf
win_crox1c.bdf ⇒ Cronyx-Courier-8.bdf
win_crox2c.bdf ⇒ Cronyx-Courier-10.bdf
win_crox3c.bdf ⇒ Cronyx-Courier-12.bdf
win_crox4h.bdf ⇒ Cronyx-Helvetica-14.bdf
win_crox5t.bdf ⇒ Cronyx-Times-18.bdf

And the same way for the rest. Include the actual font name, variation and size rather than a cryptic shortened version of both. Maybe the size should come first, since that’s often what’s most important to the user?

Finally, alt8x8.bdf should probably just be deleted.

To use any2unicode.pl you need the MAPPINGS dir from ftp://ftp.unicode.org/Public/

Based on comments in  FS#9148 , speculation is that Windows is not translating messages as if they were win1252, but something else.

Ideally, sapi_voice.vbs should be fixed so that strings are always passed in a consistent way.

One way, suggested by Taktak (Whick) is to encode the byte-value to ascii (numbers), and decode in sapi_voice.vbs. Patch for that in  FS#9148 .

This patch makes Rockbox Utility link against the system libspeex, rather than using the copy in apps/

It’s currently hard-coded and expects libspeex to be in /usr/lib - ideally it’d be smarter about this, and fall back to using the one in apps/ if there’s no libspeex to be found.

Enabling logf seems to consistently break last.fm logging for me. Tested on Sansa e200, but it seems like something that would affect all platforms.

With logf enabled, no .scrobbler.log file is created.

This patch modifies buildzip such that it will exclude a list of languages from the zipfile. Currently set to afrikaans, bulgarian, hindi, slovenscina, turkce (all languages with less than 50% translated strings), but easily changable.

Additionally, it allows to exclude languages from the Player zip, which is set to all languages that aren’t written using latin, greek or cyrillian: chinese-simp, chinese-trad, hindi, japanese, korean, thai.

Hopefully just a stop-gap measure until Daniel’s approach to  FS#8482  gets into shape.

This patch makes the build system build languages before it starts building the main binary. It also generates a file “max_language_size.h” in the build dir, containing the size of the largest language for the selected target.

Unfortunately, my build-fu is not quite strong enough to use this inside of language.h, which is the motivation for this change since it’d bring a few nice benefits:

1. We’d no longer have to keep increasing the language buffer manually each time a language grows too large.

2. The language buffer will be no larger than it needs to be. This is especially important on Archos, where this could potentially save around 10kb on the player, and less (but still significant) on the other targets.

This patch keeps a log of tracks for which no albumart file could be found. That way you could conceivably feed the log to a script which would locate missing albumart for you. This requires you to have played (actually, buffered) the file though. No effort is done to weed out duplicates, so the log could potentially grow quite large.

I haven’t bothered to make it a setting, because I’m not entirely sure it’s worth the bother, if no one but me is even remotely interested in this functionality.

The log will be in .rockbox/missing-aa.log and have the filepath, artist and albumname, separated by tabs.

This patch will make the albumart code look for albumart in the absolute path ROCKBOX_DIR/albumart/<artist>-<album><size>.bmp in addition to the current (relative) locations.

This means you can store your albumart independently of your music files.

It also adds a sanity check to fix_path_part (abort if null is encountered).

This is the current status of my work to switch to using XML for the language source-files. I basically stopped working on it in the middle of it all, and I’m not even completely sure why, right now.

What I’ve done so far:
- Created a basic structure of the file format, and documented it with an XMLschema file
- Written a tiny script that validates languages against this schema
- Written a script to convert v2 languages to XML - Started a rewritten ganlang that should end up as a drop-in replacement for the old one.

What’s missing (all in genlang):
- Deciding how to put translator notes into the source file
- Add checks for differing <source> between English and the translation
- Add header/sourcefile generation
- Fix any braindead design decisions.

See my reasons for doing this here:
http://www.rockbox.org/mail/archive/rockbox-dev-archive-2007-09/0090.shtml

This is my first stab at converting buildzip.pl to use native perl functions, rather than calling external apps (causing a pretty hefty slowdown on cygwin). So far, I’ve simply gone for the low-hanging fruit - simple cp and mv operations.

A recursive delete function should be fairly simple (to replace “rm -rf”).

The next step is probably to write a sort of “find” implementation that takes a search expression and a callback function or something like that. It doesn’t need to be a full find replacement of course. Just enough to do what buildzip needs (find and copy, I believe).

This will need a fair bit of testing of course, since we don’t want the zips to be broken.

While running a Database update with a microsd(hc) card inserted, the file browser will be unresponsive. Steps to reproduce:

1. Open the context-menu for "Database"
2. Select "Update Now"
3. Open the "Files" menu entry
4. The filebrowser will appear only when the Database update has finished (visible with statusbar turned on)

I believe it only happens with the SD card inserted, but I'm not entirely sure.

I half suspected that r15724 might have fixed it, but it doesn't seem like it has. I have no idea if it's Sansa (e200) specific.

This patch adds handling of mouseclicks to the simulator, so you can click on the buttons on the background, and those button events are then generated. It’s not possible to generate a long keypress.

I have only implemented it for the H120 sim since writing the code to detect where the mouse was clicked is rather tedious work - if anyone has an idea of how to do it better, do tell.

This is the first attempt to port the current voice scripts to perl. The benefits are/should be numerous:
- Should give a huge speedup on Cygwin
- Also gives a nice speedup in Linux & relatives.
- Should allow better use of SAPI (it might be possible to keep a single connection open)
- Perl, as horrible as it is, is probably more hacker-friendly than shell for large jobs, and should hopefully allow more people to help out.
- The configure script looks slightly nicer. And voice building is better integrated.
- voice.pl is a single file and can be run directly from the commandline. In theory all that’s needed is voicefont, genlang, english.lang and the appropriate language-file.

There are still some caveats though:
- The reason for this re-implementation was to allow TTS tweaks (phonetical replacements) - I have not added this yet.
- The script should eventually allow .talk clip generation as well. This is not yet implemented.

Given how much brouhaha the last voice change generated, I wonder if this should get a mention on either of the mailing lists? Thoughts?

The Italian language file has broken encoding. First of all it contains a BOM (byte-order marker) at the beginning of the file. Second of all, a lot of the strings contain broken characters. For example LANG_MDB_STRENGTH (and many others).

This is my first attempt at porting robotfindskitten (http://robotfindskitten.org/) to Rockbox. Right now it has a few problems:
- The screensize calculation goes wrong, and leaves a large empty area at the bottom.
- The NKI (non-kitten item) texts don’t scroll.
- General ickyness with text positioning

Wikipedia description of robotfindskitten:

robotfindskitten is a “Zen simulation,” originally written by Leonard Richardson for DOS. It is a free computer game with an ASCII interface in which the user (playing the eponymous robot and represented by a pound sign “#”) must find kitten (represented by a random character) on a field of other random characters. Walking up to items allows robot to identify them as either kitten, or any of a variety of whimsical, strange or simply random Non-Kitten Items (NKIs). It is not possible to lose

The above text is copyright by the wikipedia authors and under the GFDL license, yadda yadda.

http://en.wikipedia.org/wiki/Robotfindskitten

$ ../../tools/genlang -o
-t=h120:agc:backlight_fade:dircache:hold_button:lcd_bitmap:lcd_non-mono:lcd_flip:lcd_invert:pitchscreen:quickscreen:radio:recording:recording_swcodec:remote:remote_ticking:spdif_power:swcodec:tagcache:tc_ramcache:charging -e=english.lang dansk.lang |grep -c LANG_SET_BOOL_YES
192

Quite clearly, that is way to many instances of LANG_SET_BOOL_YES.

Assigned to bagder for good measure.

When I run genlang -u on any language, it always makes a mess out of the first phrase. The text below is what happens if I run update for english.lang itself. The same happens with other languages.

<phrase>

id: LANG_SET_BOOL_YES
desc: bool true representation
user:
<source>

### The <source> section differs from the english!
### the previously used one is commented below:
### *: “Yes” # are *identical* to the <dest> strings within the english.lang file. The
# when doing updates etc, while the <dest> ones are used when this file is
# used to simply get the full set of english strings for a particular target.
#

id: LANG_SET_BOOL_YES
user:
  *: "Yes"
</source>
<dest>
  *: "Yes"
</dest>
<voice>
  *: "Yes"
</voice>

</phrase>

This patch changes Rockbox Utility to use PNGs for the buttons. This has the advantage of allowing much nicer graphics. Right now, the image loading is implemented using external files, but it should certainly be possible to embed them in the executable.

http://wxwidgets.org/wiki/index.php/Embedding_PNG_Images Explains how to convert PNGs to C and then use them like that.

http://www.wxwidgets.org/manuals/stable/wx_wxbitmap.html#wxbitmap Says that images can either be loaded from file or a resource, but I don't know how resources work.

The next problem is I've taken icons from the Tango Desktop project. These icons are available under the Creative Commons Attribution-ShareAlike 2.5 License, which is not GPL compatible so more suitable icons should be found for embedding into the executable.

This patch makes it possible to build a debian package of rbutil. Currently it only works in Ubuntu since wxwidgets has not been built for Debian (and might not be in the foreseeable future - see http://bugs.debian.org/403237). Building the package requires the following packages installed: debhelber, fakeroot, build-essential, libwxgtk2.8-dev, devscripts, dh-make

To build the package simply execute “debuild -us -uc” in the rbutil directory. To clean up after this, execute “fakeroot debian/rules clean”.

I have created this patch using the current development version of Ubuntu (”Gutsy Gibbon”), but it ought to work on 7.04 as well, at least requiring minimal changes.

The attached .deb may or may not install on 7.04 - I have no way to check.

This exposes another problem though: that rbutil expects write permissions in the dir the binary is in. To me, this is a bug, and should be fixed to write into a dot-file in the userdir on Linux/OSX/BSD and in the user’s LocalSettings dir (I believe that’s the right one, but may be wrong) on Windows.

Morse input on the virtual keyboard is according to the manual mapped to Long down, which doesn't work - instead it just scrolls through the letters.

This patch adds an install target to the rbutil Makefile. This will be needed at the very least to build Linux packages of rbutil, and possibly also to satisfy some users’ “make; make install” reflexes.

I don’t have wx2.8 installed, so I haven’t tested this yet - maybe the $(EXT) part should be left out.