X11BASE still in use in ports

[Doug Barton]

On 06/24/2012 15:13, Eitan Adler wrote:
> On 24 June 2012 15:00, Doug Barton <dougb at freebsd.org> wrote:
>> On 06/24/2012 11:05, Eitan Adler wrote:
>>> On 24 June 2012 03:46, Doug Barton <dougb at freebsd.org> wrote:
>>>> I noticed a failure in one of my ports today while doing an upgrade, and
>>>> was embarrassed to find that it was due to my port still using X11BASE.
>>>> That led me to do a quick grep of the tree, which seems to indicate that
>>>> there are a non-zero number of uses of it which seem to be erroneous:
>>>
>>> When the patch was committed a exp-run was done.
>>
>> Which isn't even close to being a thorough treatment. The *only* way to
>> do this kind of work is with grep through the entire tree. That should
>> have been done before the variable was removed.
> 
> It wasn't the only thing that was done.

I'm trying really hard to be polite here, but let me be a little more
blunt. I don't care what you did, it wasn't sufficient. A simple grep
through the tree turned up around 30 ports that needed cleanup. Add my
port to that list as well before I fixed it last night.

The reason I'm harping on this is that it's not the first time you have
been involved in sweeping changes that were not thoroughly handled. It
would be really helpful if you could slow down a bit, and do the boring,
sloggy work of grep'ing through *every* instance of the thing you're
updating, and make sure that it's fixed.

You're a bright guy, and very enthusiastic about FreeBSD; both of which
are deeply appreciated. But while this kind of work has all the glamor
of mucking horse stalls, it makes up for it by being a giant mess if it
isn't done right.

>> FWIW, there seem to be 3 categories of problems in ak's list. First, the
>> variable isn't reached without an option being enabled (this is why
>> exp-runs are not sufficient)
> 
> I thought we fixed all of these already.

I get that you had the best of intentions. I'm not doubting that at all.

>> Second, the X11BASE was redundant, ala:
>> -I${LOCALBASE}/include -I${X11BASE}/include
> 
> I fixed the ones I saw of these, but I wasn't too concerned about
> getting them all.

*Thoroughly* removing all of the invalid uses of the old variable makes
it easier to go back through and double-check that remaining uses are
valid. It also helps to make sure that you've caught all the ones that
are actually relevant.

>> Third, it was just plain broken.
> 
> hm?

Broken. As in, without the x11base stuff being fixed, the port didn't
work properly. I gave you the examples of the fonts, look through the
diffs if you need others. (Not to mention that "failed with non-default
options" is just another category of broken.)

Doug

-- 

    This .signature sanitized for your protection