Re: Building 21.0.60 on AIX 4.3.1
Michael Sperber <sperber(a)informatik.uni-tuebingen.de> writes in
xemacs-beta(a)xemacs.org:
>>>>> "Valdis" == Valdis Kletnieks
<Valdis.Kletnieks(a)vt.edu> writes:
Valdis> [1 <text/plain; us-ascii
(7bit)>]
Valdis> On 05 Jan 1999 19:20:18 +0100, Michael Sperber [Mr. Preprocessor] said:
>> From the PROBLEMS file:
>>
>> ** AIX
>> *** On AIX 4.3, you must specify --with-dialogs=athena with configure
Valdis> Until such time as we track down the actual problem here, would it make
Valdis> sense for ./configure to default this, and to urinate all over the
Valdis> user if they attempt to overrride it, since we *know* it won't work?
I *did* figure out what the problem is in a rough way: Motif pees
all
over the data area with allocated storage.
Lovely.
My general stance is that it's a bad idea to cater to OS fuckage
in
configure, because it will alleviate pressure to get the actual bug
fixed.
How are going to force Motif bugs to get fixed? As near as I can tell
Motif has pretty much always been broken in some way or another.
But I know Steve has felt differently about this in the past.
I'll
leave it to him.
Death to Motif. Good riddance to bad rubbish. I think it's a serious
mistake defaulting to Motif support.
Valdis> I'm pretty sure that, unlike the 'xemacs -nw locks up' problem,
this
Valdis> one isn't a problem in the IBM X libraries, since I build against
Valdis> the MIT X11R6.4 and *still* see it. I haven't looked to see if it's
Valdis> an IBM xlc compiler fuckage (possibly), or if it's an actual XEmacs bug
Valdis> (but why isn't it seen elsewhere?)
No, it's Motif.
BTW: We still do need a note in PROBLEMS about AIX X libXt, don't
we?
That would be good.