Re: [2001-11-29] Freeze Update
- To: firstname.lastname@example.org
- Subject: Re: [2001-11-29] Freeze Update
- From: Richard Atterer <email@example.com>
- Date: Fri, 30 Nov 2001 23:07:13 +0100
- Message-id: <[🔎] 20011130230712.A12691@atterer.net>
- Mail-followup-to: firstname.lastname@example.org
- In-reply-to: <20011130144251.C16616@blueberry.jellybean.co.uk>
- References: <20011129010206.C22849@azure.humbug.org.au> <email@example.com> <20011129010206.C22849@azure.humbug.org.au> <firstname.lastname@example.org> <20011128183650.A29442@debian.org> <email@example.com> <firstname.lastname@example.org> <20011130144251.C16616@blueberry.jellybean.co.uk>
On Fri, Nov 30, 2001 at 02:42:52PM +0000, Jules Bean wrote:
> If all uploaded packages were built against testing libraries,
> nothing would be built against unstable libraries.
> If nothing is built against unstable libraries, then the unstable
> libraries aren't being tested.
> If they aren't tested, they should never be moved to testing!!!
Hm - surely even if a package has been built against libs in testing,
people running unstable *will* test the package's apps with any new
lib versions that are uploaded to unstable?!
> Libraries in unstable have to be tested, so packages must be built
> against them.
I'd be inclined to say instead: Library APIs must be backward
compatible under normal circumstances, so if a new lib version breaks
apps compiled against old lib versions, this is bound to be a bug in
the lib, not the app. Consequently, it is OK if the app has already
moved to testing by the time the problem with the new lib turns up.
|_) /| Richard Atterer | CS student at the Technische | GnuPG key:
| \/¯| http://atterer.net | Universität München, Germany | 0x888354F7
¯ ´` ¯