Bug#916933: RFS: libgsm/1.0.18-1 [ITA]
Hi Adam, On Fri, Dec 21, 2018 at 11:27 AM Adam Borowski wrote: > > So please give me something to re-test and upload. Please look again on Mentors. https://mentors.debian.net/package/libgsm > I'd be happy with either a hard break or a compat link -- it's up to you and > your upstream wrt what is preferred. Upstream ships the sole header in /usr/include/gsm.h. Fedora and its derivatives, on the other hand, ship the header in /usr/include/gsm/gsm.h and set a symlink at /usr/include/gsm.h. In Debian, we also use those locations (it's the same file). If you have no objections, I would like to proceed with Fedora's model. >From my experience, most software will look at either location, or sometimes at both. (My two packages mediastreamer2 and svxlink do.) Also, the codec is often optional, as you already noted, and may be omitted unless found. The upcoming release deadline is another factor. Most people are busy with other things. In summary: The library is over twenty-five years old; I would prefer to accept the status quo (while fixing #882176). On my system with the version in Mentors, these consumers built successfully via sbuild: asterisk freerdp2 gmerlin-avdecoder mediastreamer2 twinkle
Bug#916933: RFS: libgsm/1.0.18-1 [ITA]
Hi Adam, On Fri, Dec 21, 2018 at 11:27 AM Adam Borowski wrote: > > Hi; > the package has been for some reason removed from mentors, despite no one > uploading anything. Could you tell me if there's a version you'd want in? > I removed the package from Mentors in order to upload another one, but made a mistake and aborted with ^C. After that, Mentors refused additional uploads with 403 Forbidden. In the past, I have had success after waiting for Mentors to clean up the incoming queue. This morning I saw the RFS bug was closed automatically as a result. It was an inadvertent chain of events. > I'd be happy with either a hard break or a compat link -- it's up to you > and > your upstream wrt what is preferred. I believe we know enough; thanks for > your work so far. > The software is from 1992. I may contact them to get guidance. > So please give me something to re-test and upload. > I will be back with a new version in a few days. Thank you for your cooperation so far!
Bug#916933: RFS: libgsm/1.0.18-1 [ITA]
On Fri, Dec 21, 2018 at 05:23:19AM +0100, Adam Borowski wrote: > On Thu, Dec 20, 2018 at 01:14:01PM -0800, Felix Lechner wrote: > > On Thu, Dec 20, 2018 at 12:10 PM Adam Borowski wrote: > > > audio_gsm.c:29:11: fatal error: GSM610/gsm.h: No such file or directory > > > # include "GSM610/gsm.h" > > > > Thank you for taking a look. The errors you found are due to confusion > > about the include path. > > Full results of the rebuilds: > > freerdp2_amd64.build:Status: attempted > gmerlin-avdecoder_amd64.build:Status: attempted > mediastreamer2_amd64.build:Status: attempted > twinkle_amd64.build:Status: attempted > > The previous version shipped gsm.h in two locations: It was in > > /usr/include, and again in /usr/include/gsm---together with a bunch of > > probably private headers that gave rise to #882176. > > Either way, I uploaded a new version to Mentors that again ships the > > same file in both locations. Would you please try again? Hi; the package has been for some reason removed from mentors, despite no one uploading anything. Could you tell me if there's a version you'd want in? I'd be happy with either a hard break or a compat link -- it's up to you and your upstream wrt what is preferred. I believe we know enough; thanks for your work so far. So please give me something to re-test and upload. Meow! -- ⢀⣴⠾⠻⢶⣦⠀ ⣾⠁⢠⠒⠀⣿⡁ Ivan was a worldly man: born in St. Petersburg, raised in ⢿⡄⠘⠷⠚⠋⠀ Petrograd, lived most of his life in Leningrad, then returned ⠈⠳⣄ to the city of his birth to die.
Bug#916933: RFS: libgsm/1.0.18-1 [ITA]
On Thu, Dec 20, 2018 at 01:14:01PM -0800, Felix Lechner wrote: > On Thu, Dec 20, 2018 at 12:10 PM Adam Borowski wrote: > > > > audio_gsm.c:29:11: fatal error: GSM610/gsm.h: No such file or directory > > # include "GSM610/gsm.h" > > Thank you for taking a look. The errors you found are due to confusion > about the include path. Full results of the rebuilds: asterisk_amd64.build:Status: given-back baresip_amd64.build:Status: successful blender_amd64.build:Status: successful ffmpeg_amd64.build:Status: successful flightgear_amd64.build:Status: successful freerdp2_amd64.build:Status: attempted gmerlin-avdecoder_amd64.build:Status: attempted gnuradio_amd64.build:Status: successful gst-plugins-bad1.0_amd64.build:Status: successful kino_amd64.build:Status: successful linphone_amd64.build:Status: successful mangler_amd64.build:Status: given-back mediastreamer2_amd64.build:Status: attempted mplayer_amd64.build:Status: successful pjproject_amd64.build:Status: successful ring_amd64.build:Status: successful rplay_amd64.build:Status: successful sipxtapi_amd64.build:Status: successful sox_amd64.build:Status: successful svxlink_amd64.build:Status: successful swh-lv2_amd64.build:Status: successful swh-plugins_amd64.build:Status: successful twinkle_amd64.build:Status: attempted vlc_amd64.build:Status: successful wine-development_amd64.build:Status: given-back wine_amd64.build:Status: given-back http://ix.io/1wu4 http://ix.io/1wu3 Breakage looks same. > The previous version shipped gsm.h in two locations: It was in > /usr/include, and again in /usr/include/gsm---together with a bunch of > probably private headers that gave rise to #882176. When trying to fix > the latter, I settled for /usr/include/gsm.h since the file appeared > to be the sole header (although the reporting party's patch kept it in > /usr/include/gsm). As a side note, I had seen #include in > a package of mine but thought it was conditional on autoconf; some of > the packages you tested may also look for the header in both places. > Either way, I uploaded a new version to Mentors that again ships the > same file in both locations. Would you please try again? > > If you feel strongly about it, we can also file bugs against packages > that use the library but do not look for the header in your favorite > location. I have no preference. I'm slightly biased towards making a hard break rather than a compat hack that'd linger -- the former is more immediate work but is far cleaner. On the other hand, packages might silently _succeed_ to build while skipping gsm support, and there's little time to find such bugs before Buster's freeze. Do you happen to know what upstream does? Meow! -- ⢀⣴⠾⠻⢶⣦⠀ ⣾⠁⢠⠒⠀⣿⡁ Ivan was a worldly man: born in St. Petersburg, raised in ⢿⡄⠘⠷⠚⠋⠀ Petrograd, lived most of his life in Leningrad, then returned ⠈⠳⣄ to the city of his birth to die.
Bug#916933: RFS: libgsm/1.0.18-1 [ITA]
Hi Adam, On Thu, Dec 20, 2018 at 12:10 PM Adam Borowski wrote: > > audio_gsm.c:29:11: fatal error: GSM610/gsm.h: No such file or directory > # include "GSM610/gsm.h" Thank you for taking a look. The errors you found are due to confusion about the include path. The previous version shipped gsm.h in two locations: It was in /usr/include, and again in /usr/include/gsm---together with a bunch of probably private headers that gave rise to #882176. When trying to fix the latter, I settled for /usr/include/gsm.h since the file appeared to be the sole header (although the reporting party's patch kept it in /usr/include/gsm). As a side note, I had seen #include in a package of mine but thought it was conditional on autoconf; some of the packages you tested may also look for the header in both places. Either way, I uploaded a new version to Mentors that again ships the same file in both locations. Would you please try again? If you feel strongly about it, we can also file bugs against packages that use the library but do not look for the header in your favorite location. I have no preference. Roar & thank you!
Bug#916933: RFS: libgsm/1.0.18-1 [ITA]
On Thu, Dec 20, 2018 at 08:36:07AM -0800, Felix Lechner wrote: > * Package name: libgsm >Version : 1.0.18-1 > dget -x > https://mentors.debian.net/debian/pool/main/libg/libgsm/libgsm_1.0.18-1.dsc > Changes since the last upload: > > * New upstream version > * New maintainer (Closes: #891760) > * Only ship gsm.h (Closes: #882176) > * Migrated rules to dh(1) > * Updated patches (most of 05 was accepted upstream) > * Bumped shared object version in 01 patch > * Made copyright file dep5 machine readable > * Added README.Debian about (non-)patent status > * Updated upstream URL in watch file > * Turned on migration to automatic debug packages > * Removed old debug package from control > * Removed libgsm1-dbg.install > * Added homepage field to control > * Removed trailing whitespace from changelog > * Removed trailing whitespace from control > * Removed ancient Conflicts: from control > * Corrected two misspellings in man via 06 patch > * Removed duplicate binary control field for section > * Set Build-Depends: debhelper (>= 11) > * Set compat to 11 > * Updated Standards-Version: 4.2.1 Alas, while your changes to the package itself look good, at least some of reverse-depends don't build anymore. So far, I tried the following: asterisk_amd64.build:Status: given-back baresip_amd64.build:Status: successful blender_amd64.build:Status: successful ffmpeg_amd64.build:Status: successful flightgear_amd64.build:Status: successful freerdp2_amd64.build:Status: attempted gmerlin-avdecoder_amd64.build:Status: attempted gnuradio_amd64.build:Status: successful gst-plugins-bad1.0_amd64.build:Status: successful kino_amd64.build:Status: successful freerdp2: http://ix.io/1wsw gmerlin-avdecoder: http://ix.io/1wsx CMake Error at /usr/share/cmake-3.13/Modules/FindPackageHandleStandardArgs.cmake:137 (message): Could NOT find GSM (missing: GSM_INCLUDE_DIR) audio_gsm.c:29:11: fatal error: GSM610/gsm.h: No such file or directory # include "GSM610/gsm.h" Thus, you need to investigate and decide what to do. Meow! -- ⢀⣴⠾⠻⢶⣦⠀ ⣾⠁⢠⠒⠀⣿⡁ Ivan was a worldly man: born in St. Petersburg, raised in ⢿⡄⠘⠷⠚⠋⠀ Petrograd, lived most of his life in Leningrad, then returned ⠈⠳⣄ to the city of his birth to die.
Bug#916933: RFS: libgsm/1.0.18-1 [ITA]
Package: sponsorship-requests Severity: normal X-Debbugs-CC: Mattia Rizzolo Dear mentors, I am looking for a sponsor for my package "libgsm" * Package name: libgsm Version : 1.0.18-1 Upstream Author : Jutta Degener * URL : http://www.quut.com/gsm/ * License : TU-Berlin-2.0 Section : libs It builds those binary packages: libgsm-tools - User binaries for a GSM speech compressor libgsm1- Shared libraries for GSM speech compressor libgsm1-dev - Development libraries for a GSM speech compressor To access further information about this package, please visit the following URL: https://mentors.debian.net/package/libgsm Alternatively, one can download the package with dget using this command: dget -x https://mentors.debian.net/debian/pool/main/libg/libgsm/libgsm_1.0.18-1.dsc The last upload for this library was in 2012. Since two of my packages use it, I decided to adopt it. Changes since the last upload: * New upstream version * New maintainer (Closes: #891760) * Only ship gsm.h (Closes: #882176) * Migrated rules to dh(1) * Updated patches (most of 05 was accepted upstream) * Bumped shared object version in 01 patch * Made copyright file dep5 machine readable * Added README.Debian about (non-)patent status * Updated upstream URL in watch file * Turned on migration to automatic debug packages * Removed old debug package from control * Removed libgsm1-dbg.install * Added homepage field to control * Removed trailing whitespace from changelog * Removed trailing whitespace from control * Removed ancient Conflicts: from control * Corrected two misspellings in man via 06 patch * Removed duplicate binary control field for section * Set Build-Depends: debhelper (>= 11) * Set compat to 11 * Updated Standards-Version: 4.2.1 Regards, Felix Lechner
