Re: [SM-CVS] CVS: squirrelmail/config configtest.php,1.1.2.1, 1.1.2.2

[Marc G. K. <ma...@it...>]

Erin Schnabel zei:
> Chris Hilts said:
>>> I deliberately put this in stable, because it's a self-contained
>>> feature
>>
>> Is it a bugfix?  No?  Devel.
>
> ==> He has a point, Thijs. This should have been checked in/verified
> against devel and then moved back to stable. You, as stable lead, should
> know this.
OK here my opinion about configtest.php

We reviewed it during an IRC session and the configtest script is not part
of the core, it's an add-on that doesn't interfere with SM-core.
I can imagine that this script is very usefull for tracking down
misconfigurations of certain users and therefor is a usefull tool for the
STABLE branche. As Stable-lead Thijs and Jon put a lot of efforts into
helping users with a misconfigured config.php. That's why I think
configtest.php belongs in the STABLE branch-and not in the devel-branche.
It helps them tracking down thos misconfigurations a lot quicker.

>
>>
>>> It can even enhance the perception of stability to the user in the
>>> sense
>>
>> It's new code.  You committed it full of errors and inconsistencies.
>> It's
>> had no peer review. It's a new feature.  Good god, why is this even a
>> debate?
>
> Chris, it did have perfunctory peer review. Thijs mentioned it, and I
> looked at it breifly, and liked the function. However, *I* assumed it
> would get commited to devel, and that it wouldn't be committed at all
> until better tested, as Thijs himself said..
>
OK, a case of misinterpretation. We have dealed with that before :-)

I don't know if Thijs discussed it with Jon but if he have I think the
devel-team should have confidence in the STABLE commits.

>>
>>> file, just to be sure. But there's no real harm in some dude running
>>> the
>>> script without the administrator knowing it; it doens't give
>>> information
>>> that isn't otherwise easily accessible.
>>
>> This is debatable, and really beside the main point I was trying to
>> make.
>> If this makes it to devel, I'll probably throw a config option somehow
>> to
>> enable/disable the config tester.
>>
>> 1.4.0 has been released as the official "stable".  That means we don't
>> add
>> new features or code that isn't a bugfix.  New code, new features belong
>> in 1.5.x.
>
> I agree. And even if it at some later point the devel/stable streams are
> so dissimilar that new things have to be worked into stable, that is not
> the case right now.
>
Maybe this is interesting to discuss about the next time. We just passed a
long period of feature freeze to make 1.4 STABLE and get it out of the
door. In the past there were a lot of new feature-adds to STABLE and
nobody complains about that. Maybe the situation is changed because of
devel-branche but things like register globals = off was initially added
to the STABLE core and at a later time backported to DEVEL.

On the other hand you are right that at this moment DEVEL and STABLE are
pretty simular so at this time things can be tested in the DEVEL branche.

So my sugestion: let's talk about what is and isn't allowed to add to
STABLE so we do not have to deal again with "grey" areas.

>
> Erin
> (ebullient)
>

Marc
 (stekkel)   <-------- (scooped this from Erin)