Hi Marc,
Marc Espie wrote on Sat, Jul 28, 2018 at 12:33:22AM +0200:
> On Fri, Jul 27, 2018 at 03:45:50PM +0200, Ingo Schwarze wrote:
>> espie@ and ajacoutot@ would seem like a good team for deciding that
>> to me, taking input from other porters working on infrastructure
>> into account.
> I'm actually for keeping LOCALBASE support.
Then, consider providing an OK for solene's diff, or telling her
how to improve it. :-)
>>> bsd.port.mk(5) says
>>>
>>> BUGS AND LIMITATIONS
>>> LOCALBASE, X11BASE, SYSCONFDIR and PREFIX are not heeded
>>> consistently. Most of the ports tree will probably fall apart
>>> if one tries to build/use stuff elsewhere.
I took that statement at face value, given that Antoine appears to
be convinced that it is accurate.
>> To me, that sounds like a very strong argument to remove support
>> for these features as suggested by ajas 'let us just hardcode
>> /usr/local and be done with it' gang.
>>
>> I mean, seriously, since when does OpenBSD indulge in providing bells
>> and whistles that are not even *supposed* to actually work? Besides,
>> i have often heard even core porters groan that some aspects of the
>> ports infrastructure are hard to master - most of that is certainly
>> unavoidable because porting is a complex business, but when aiming
>> for KISS, every small bit helps, and this does seem like an obvious
>> chance to get rid of some complexity.
> Technically, fixing that is waays simpler than it was two years ago.
> Building stuff chroot'd by default means the layout of the filesystem
> is more flexible, and I can very easily decide to put stuff outside of
> /usr/local in my chroot if I want.
>
> So yeah, it actually is on my todo list.
> The current hooks are not complicated at all, really.
Hmm, i'm not quite sure how to reconcile "easy to fix" and "most of the
ports tree will fall apart when trying to use it" - but if you think
you can fix it...
If the two chief maintainers disagree about the usefulness of an
established feature, we don't usually remove it.
Sure, i see your irony, and i think the documentation of LOCALBASE
and PREFIX in bsd.port.mk(5) is actually clear and concise. I think
this particular case is not a problem of incomplete or unclear
documentation. The problem probably is that porters don't remember
how these variables work because the feature these variables implement
is almost never useful for anybody, yet everybody has to always use
the variables, and to use them correctly. Also, correct use of the
variables is practically never tested, and testing it might even be
a wasteful make-work project.
You can choose to not learn how "man -k Ev=" and ":t" in less work
and yet use man(1) all day without harming anybody. That's your
choice. But you cannot choose to not learn LOCALBASE and PREFIX
and yet commit ports all day. You have to learn them even if you
never use them personally. I'm sure people honestly try. And yet,
according to the manual page, the outcome is that even after the
review implied by the OK you need for a commit, significant parts
of the ports tree appear to be incorrect. That doesn't sound like a
simple thing to me.
Besides, bsd.port.mk(5) contains 108 targets and 276 variables, and
many of them are required for doing proper porting. Apropos has
13 options and 41 macro keys and using any of them is totally
optional.
But again, if there is no consensus among the maintainers that a
feature is detrimental, it will stay and hopefully be improved
as time permits.
Yours,
Ingo
No comments:
Post a Comment