Re: Bug#1052597: RFS: libkysdk-base/2.2.0.1-1 -- common files for kylin sdk base library
- To: debian-mentors@lists.debian.org
- Subject: Re: Bug#1052597: RFS: libkysdk-base/2.2.0.1-1 -- common files for kylin sdk base library
- From: Sune Vuorela <nospam@vuorela.dk>
- Date: Tue, 19 Dec 2023 08:05:17 -0000 (UTC)
- Message-id: <[🔎] ulrirt$56a$1@ciao.gmane.io>
- References: <tencent_7008AE86597DB5175D1ED5A8@qq.com> <tencent_546307164CCC16EC2BCE08AE@qq.com> <tencent_7008AE86597DB5175D1ED5A8@qq.com> <ZYCdz1GwDRwA0mIJ__22984.286448785$1702928028$gmane$org@isildor2.loewenhoehle.ip>
On 2023-12-18, Tobias Frost <tobi@debian.org> wrote:
> - Updating the SONAME of a library requires this procedure to be followed:
> https://wiki.debian.org/Teams/ReleaseTeam/Transitions
> Comparing the symbols file does not make it obvious why you are
> bumping SONAME, but I did not check with abi-complicance-checker...
> Can you fill me in why you bump the soname?
>
> - the breaks/replaces version seems odd, as it is a binnmu version.
> You likely want (<< 2.2.0.1-1~), though I am not sure why you think
> you'll need the Break/Replace? Can you exand?
>
> - you could use d/clean instead of overriding dh_clean
>
> - for the install files, for multiarch, a cleaner way would be to write
> /usr/lib/${DEB_HOST_MULTIARCH}/… instead of /usr/lib/*/…
Why does the build system force an ancient c++ standard ? It seems to
mostly get in the way and there are rarely good reasons to do it.
Why does it build with -O1 in release mode ?
How do you plan on dealing with just a SONAME break in one of the
libraries?
Why is nlohman-json embedded and not used from system?
- Also just a quick review.
/Sune
Reply to: