On Thu, Mar 13, 2025 at 10:50:36PM +0100, Страхиња Радић wrote:
> IMHO the latest additions to POSIX are nonportable to programs and
> systems using an earlier versions of the standard,
This is a very good point and I would readily accept it as a refutation
to my patch. I have found that all _modern_ make(1) implementations conform
to the standard and would process this Makefile (following the patch)
correctly, but the previous Makefile (without my patch) would be
recognized by pretty much any make(1) every written (hyperbole).

Personally, I would say that programs not conforming to the standard are
broken and should not be catered to, but specifically for something
like sbase, working with older systems (even bootstrapping, etc) is
pretty important and shouldn't be prevented for petty reasons (like I
said in the patch, this isn't a totally necessary/important change, but
I did it for fun. I also find this discussion fun, hence why I am
continuing :p).

> and the suckless movement should fix the previous version of POSIX
by `fix' POSIX, are we referring to restricting ourselves to an older
version (hopefully one that is chosen because it is the latest version
which is functionally implemented on important systems/in use by a
significant enough number of people), cherry picking parts of various
standards that we like to create a Franken-standard, or brewing up our
own standard based on some version of POSIX, plus extensions we thing
are nice.

I think the first option would be acceptable, but the other two are
worse than simply not conforming (they basically amount to `do whatever
the hell we want', which means that same thing as `nonconforming', even
if we wrap it up in nice logos).

> similar to fixing an earlier version of the C standard (C99).
I'm terribly sorry if I'm showing my ass here by not knowing exactly
what this refers to, but the thing that immediately comes to mind is the
Plan9 dialect of C. While I do like that a lot, I wouldn't really use it
because it isn't supported by the most common compilers and I don't want
to carry a compiler around just to use some language features I like :/.
 
> And here we arrive again at the question of what exact version of POSIX
> should the suckless movement adopt.
-- 
S

Reply via email to