We build Tcl 8.4 at work, among many other things. It fails to build on one HPUX machine when built from the overnight build infrastructure; when built manually it worked fine. The error was a parse error from sed, somewhere in the depths of config.status. The actual offending command was identical though there was some slight difference further up the script (just path names). Moving the command itself further up the script identified a point (not obviously related to those differences) where it all started behaving (but that's no good as a solution of course as it's automatically generated). I'd already managed to eliminate environmental differences.
I gave up and installed GNU sed and it just worked.
Back in 1994, I experienced the joys of IRIX's C compiler segfaulting on a conforming ANSI C program. When I transformed it into a conforming pcc program, that made it segfault, too.
I hope their compiler has got a lot better, since.
There was also the lovely trick they used to do of shipping with undocumented, unpassworded shell accounts.
Mmmm. More recently, 1999 or so, we got an IRIX machine from SGI which, out of the box, offered login access to the unpassworded "guest" account over telnet.
Argh, poor you. I (really, used to) run some HP-UX 10.x machines and they really were something special. sed was indeed an issue, except much regular GNU stuff had a hard time compiling too. ISTR having to fix broken C in the HP-UX /usr/include/ header files to get some GNU stuff working. g++ was a real pain IIRC.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
(no subject)
Date: 2003-07-07 04:00 pm (UTC)(no subject)
Date: 2003-07-08 12:55 am (UTC)We build Tcl 8.4 at work, among many other things. It fails to build on one HPUX machine when built from the overnight build infrastructure; when built manually it worked fine. The error was a parse error from sed, somewhere in the depths of config.status. The actual offending command was identical though there was some slight difference further up the script (just path names). Moving the command itself further up the script identified a point (not obviously related to those differences) where it all started behaving (but that's no good as a solution of course as it's automatically generated). I'd already managed to eliminate environmental differences.
I gave up and installed GNU sed and it just worked.
(no subject)
Date: 2003-07-07 04:12 pm (UTC)-m-
(no subject)
Date: 2003-07-08 12:55 am (UTC)(no subject)
Date: 2003-07-08 03:14 am (UTC)I hope their compiler has got a lot better, since.
There was also the lovely trick they used to do of shipping with undocumented, unpassworded shell accounts.
IRIX - nice filesystem, shame about... well, everything else, really.
(no subject)
Date: 2003-07-08 10:23 am (UTC)(no subject)
Date: 2003-07-08 10:26 am (UTC)