Hi Ian,
On 19-06-18 23:23, Ian Jackson wrote:
> Paul Gevers writes ("Re: DEP-8 pseudo-restriction
> "hint-dpkg-testsuite-triggers""):
>>>> Considering your description above I give you the following alternative
>>>> suggestion: make a gnupg1 specific test, where you install gnupg instead
>>>> of gnupg2 and verify that it works also that way. For a while I was
>>>> testing my package dbconfig-common with both MariaDB and MySQL servers
>>>> (neither of which are my (indirect) dependencies as the server may run
>>>> on a different host).
>>>
>>> What, do that and add the gnupg direct dependency to all the other
>>> tests ?
>>
>> No, have just one specific test for gnupg1, and only add it there.
>
> I think we must be talking at cross purposes. I'm going to try to
> explain again, to give you another chance to convince me I'm going
> about this the wrong way. So:
>
> gnupg2 is the current default in Debian. dgit is supposed to work
> with both gnupg1 and gnupg2. It calls gnupg both directly and via
> debsign. In practice there is no difficulty with gnupg1. I worry,
> though, about possible regressions in interaction with gnupg2 and
> gpgv.Ah, I thought you were worried about gnupg1's behavior as gnupg2 is the default. > Almost every test makes and verifies signatures, using a stunt set of > keys. When testing switched to gnupg2, I discovered (sadly too late > to make a fuss about it, by the time I had figured out what was going > on) the gnupg2 startup bugs. I have been adding increasingly horrific > workarounds to the test suite. > > I want an early warning if things get worse. I think the Debian > gnupg2 maintainers would appreciate that too. (Upstream's attitude is > ... more mixed.) So I need my tests rerun in buster, with gnupg2, > when gnupg2 is updated. > > Having a test run with gnupg1 might be a bonus but is not really > important, since gnupg1 works and, realistically, probably isn't going > to break. I don't know the design of your tests and I doubt it would matter, but reading what you said above I would probably force one of all test cases to use gnupg1 and another one to force gnupg2. That is what we did for debian-edu. There is a whole (2 or 3 dimensional IIRC) matrix of combinations that ideally one would test all, but most of the time that doesn't make sense as that would be a huge waste of the infrastructures time. So the debian-edu team picked points in the matrix to cover at least all points on all axes at least once and I believe there are some additional combinations. Paul
signature.asc
Description: OpenPGP digital signature
