Re: release 3.0.2

[Jaroslav Hajek]

On Tue, Aug 19, 2008 at 3:52 PM, Sergei Steshenko <address@hidden> wrote:
>
>
>
> --- On Tue, 8/19/08, address@hidden <address@hidden> wrote:
>
>> From: address@hidden <address@hidden>
>> Subject: Re: release 3.0.2
>> To: address@hidden
>> Cc: "Jaroslav Hajek" <address@hidden>, address@hidden
>> Date: Tuesday, August 19, 2008, 5:41 AM
>> Quoting Sergei Steshenko <address@hidden>:
>> > I do not think it is decent to make a release of
>> whatever SW product if
>> > it is known that built-in test suite fails.
>>
>> I got very early out of bed this morning, so I'm not in
>> the best mood.
>> So perhaps I'm over-reacting, but I don't think
>> it's fair to call the
>> release manager indecent due to a bug in a release.
>>
>> The process was:
>>    1) Octave 3.0.2 was released.
>>    2) A bug was found.
>> And now you conclude that the release should be canceled?
>> Bugs are
>> found every day, and fixes are made every day. That
>> doesn't mean
>> releases should be made every day.
>>
>> Søren
>
> ???
>
> I didn't call the release manager indecent.
>
> I probably do not understand the release procedure of 'octave'.
>

Well, I guess there is no written standard here. We can certainly
discuss any part of the procedure.

>
> My expectations of a release procedure, as I wrote earlier, is that a
> release is not made while there are failures in built-in test suite.
>
> Standard
>
> ./configure
> make
> make check
>
> sequence should be run before any release.

Yes, I have done this. I think I must have overlooked this particular
failure, or just forgot to commit the single character change, because
I think the check was clear at least on my x86/Linux configuration.

>
> I.e. built-in test suite should cleanly run before a release; if it doesn't
> run cleanly, the release is not made/announced.
>

This is where we disagree. There might be failing tests which we don't
know how to fix, or there might be an ongoing dispute.

> If the release manager ran the sequence, and it showed no bugs, and other
> builders see some bugs, then there's something wrong either in other
> builders' setup or in release manager's setup or in tests.
>

Exactly. But what to do then?

> Regards,
>  Sergei.
>
>
>
>
>
>



-- 
RNDr. Jaroslav Hajek
computing expert
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz