Re: git: 00e15405660f - main - textproc/html2text: Update CONFLICTS

Am 28.10.21 um 04:33 schrieb Alexey Dokuchaev:
> On Wed, Oct 27, 2021 at 08:25:13PM +0200, Stefan Esser wrote:
>> Am 27.10.21 um 11:00 schrieb Alexey Dokuchaev:
>>>> ...
>>>> -CONFLICTS=	py27-html2text-[0-9]*
>>>> +CONFLICTS=	py*-html2text-[0-9]*
>>>
>>> I recal, last time I've tried to specify a conflict without the -[0-9]*
>>> suffix and it worked as expected.  Are those really necessary in default
>>> "all versions" case?
>>
>> And quite a few CONFLICTS specifications should actually be
>> CONFLICTS_INSTALL. But it takes a lot of work to check whether there
>> really is a build conflict for each of the combinations...
> 
> I'm not sure if CONFLICTS alone is justified, and why can't we always
> use correct CONFLICTS_$TYPE.  Checking if CONFLICTS should really be
> CONFLICTS_INSTALL could be automated, or so it seems at least for cases
> when prerequisite are met, by comparing pkg-plists of ports in question.
> 
> CONFLICTS_BUILD are tougher; whenever I added one, it was as a result
> of stumbling across local build problem and wondering why it does not
> happen on the cluster, i.e. in pristine environment.
> 
> Overall, many good points Stefan, nice write up.  So that even if you
> won't pick up on this task, someone might now that you've made it less
> of an uncharted territory.

Well, I went ahead and have generated a list of ports that have the
redundant "-[0-9]*" version suffix in CONFLICTS patterns.

As of today, these are 388 ports in total.

It was easy to write a script that removes those suffices and I have
generated a review:

	https://reviews.freebsd.org/D32703

If that review was accepted by portmgr@, it would clean up the ports
tree with regards to this issue.

Else I can send patches to individual port maintainers and ask them
for approval or let them apply the patches to their ports.

This does not address CONFLICTS vs. CONFLICTS_INSTALL or other aspects,
obviously. But I think it is still a reasonable simplification and does
prevent permeation of -[0-9]* suffices when a new port is created based
on an existing one with this version wildcard present ...

Regards, STefan