Bug#858552: Acknowledgement (shairport-sync: shairport-sync user and group are not removed when package is purged)
Control: tags -1 wontfix Control: severity wishlist On 24/03/17 15:59, Denys Berkovskyy wrote: > Hi, > > Please find attached a path which fixes the issue. Since I was fixing issue > in git I have as well pushed changes to github and attached pull-request. To > test the changes I have made a package with changes applied to new upstream > version and uploaded it to mentors.debian.org: > https://mentors.debian.net/package/shairport-sync > > Alternatively you can download the package from mentors with dget: > dget -x > https://mentors.debian.net/debian/pool/main/s/shairport-sync/shairport-sync_3.0.2-0.1.dsc > > > I am new to the Debian packaging and I am not sure if I am doing the things > correctly. Could you let me know what should I do in the future. Is it > preferable to create and send patched on BTS, or doing it in git repository > and sending pull-requests is preferred way? And if creating packages with > changes applied is any use to maintainers? I would as well appreciate any > other comments you have. Hi Denys, First of all, sorry for leaving your bug report hanging for so long despite you being ultra-helpful and including a patch. I don't think the package should remove its user or group when it is removed or purged. The reason is twofold: 1- should there be any files leftover on your system owned by said user they will now show up only with their UID/GID numbers, and 2- if you then install another package the UID/GID will be recycled and they will be owned by the new user. shairport-sync doesn't install any files itself, and the only leftover items will likely be under /run, but I can't guarantee that users won't chown/chgrp files elsewhere on their filesystem, leading to confusion and even potential security issues. I had a scout around other common packages, and none that I looked at remove their users/groups. Aside from that your patch looks perfectly valid and it was very helpful that you went out of your way to learn how to do this. Please don't be discouraged by me not applying your patch! Attaching the patch to the bug and uploading to mentors were both very good things to do. Debian developers have a tendency not to use GitHub for a plethora of reasons. I don't personally mind GitHub at all and use it for various things, but I have used Debian's Alioth project system to host my Debian packages. All this means is that pushing to GitHub and doing the pull request was unnecessary, even if some people might find it helpful. Thanks again, Chris -- Chris Boot bo...@debian.org GPG: 8467 53CB 1921 3142 C56D C918 F5C8 3C05 D9CE signature.asc Description: OpenPGP digital signature
Bug#858552: Acknowledgement (shairport-sync: shairport-sync user and group are not removed when package is purged)
Hi, Please find attached a path which fixes the issue. Since I was fixing issue in git I have as well pushed changes to github and attached pull-request. To test the changes I have made a package with changes applied to new upstream version and uploaded it to mentors.debian.org: https://mentors.debian.net/package/shairport-sync Alternatively you can download the package from mentors with dget: dget -x https://mentors.debian.net/debian/pool/main/s/shairport-sync/shairport-sync_3.0.2-0.1.dsc I am new to the Debian packaging and I am not sure if I am doing the things correctly. Could you let me know what should I do in the future. Is it preferable to create and send patched on BTS, or doing it in git repository and sending pull-requests is preferred way? And if creating packages with changes applied is any use to maintainers? I would as well appreciate any other comments you have. Regards Denys Berkovskyy 0001-Add-shairport-sync-user-and-group-removal-to-postrm.patch Description: Binary data The following changes since commit ab43e10e51e3da4296bea600d8d9e53a7179f1b0: Upload 2.8.6-1 to unstable (2016-11-11 15:41:09 +) are available in the git repository at: https://github.com/berkovskyy/debian-shairport-sync.git master for you to fetch changes up to 0aa521227d6dc9f240e736aa88b5e6fe9fd955cc: Upload 3.0.2-0.1 to unstable (2017-03-23 11:39:23 +) Denys Berkovskyy (5): Update changelog for 3.0.2 Merge tag 'upstream/3.0.2' Refresh patches Add shairport-sync user and group removal to postrm script in case of packet purging. Upload 3.0.2-0.1 to unstable Jief (4): constant 0x1 should be 0x1LL to avoid warning Using Macro instead of LL to make it portable. Using macro instead of LL to make it portable clock_drift could be left uninitialized. Jörg Krause (2): mdns: avahi: set AVAHI_CLIENT_NO_FAIL in avahi_client_new() mdns: avahi: handle avahi client failure Mike Brady (157): Add ability to set output format and output rate fix tiny syntax error Great things happening... Begin to add in calculations for expanded back end Fix a comment typo. Start the great trek Make it compile -- untested Merge support of apple's ALAC decoder Make the decoder choice a bit more general, highlight names of input format parameters Small improvements Apple ALAC decoder not working -- magic cookie contents are faulty. Begin to add support for monotonic 64-bit timestamps and frame sequence numbers Add more support for monotonic 64-bit timestamps and frame sequence numbers Enable 64-bit monotonic rtptime calculations Fix a few little bugs associated with outputing at a higher rate Add code to change format from 16 to 24 and 32 bit and to change sample rates Adding rate and format transformation stuff Working reliably with all 16-bit speeds Working reliably with all 16-bit speeds Change over to times as opposed to frames for setting timing options Change over to times as opposed to frames for setting timing options Fix small bug introduced moving over to time-based settings Stop multiplying monotonic timestamps by the output to input sampling ratio Fix silence bug and hack code to avoid the stuff procedures Fix silence bug and hack code to avoid the stuff procedures Preliminary support for 32-bit stuffing Preliminary support for 32-bit stuffing Add a fancy 64-bit pseudo random number generator -- thanks to its developers Add a fancy 64-bit pseudo random number generator -- thanks to its developers Remove prng initialisation code -- it's not needed Remove mersenne generator -- too slow, try another one Get dither mask working right Begin to add full-scale post processing. Does not compile. New backend for converting 32-bit "native" samples to output format ready for testing New backend for converting 32-bit "native" samples to output format ready for testing Some bug fixes and hacks. Not working. Add initial support for U8 and S8 formats Beginning to work, also added S8 and U8 support as VMWare Fusion and the Pi both offer them Tidy up with a clang-format Soxr interpolation working. Fix 8-bit dithering, fix software volume control format player.c Soxr interpolation working. Dithering seems to work. Software volume seems to work trying to get a sane build numbering system. 'd' for 'development'... Make endian-independent. Allow users to specify only U8/S8/S16/S24/S32 Make endian-independent. Allow users to specify only U8/S8/S16/S24/S32 Fix error checking of alsa latency and buffer legth settings. Get different playback modes working and update the configuration