Last night I tried to use "distcc" (2.17) to build Mozilla Firefox 0.9.3 on Solaris 9 using the Sun Forte compilers.
It died in exactly the same way as what happened to someone else on the list back on March 10th as described in the list archives:
http://www.mail-archive.com/[EMAIL PROTECTED]/msg01891.html
At the time, Martin said
"Sun CC is not fully supported by distcc at the moment. Perhaps you can use gcc, if your source works with that."
That was 5 months ago. Has any work been done on this front since?
I was really jazzed about "distcc" - it takes 38 minutes to build Firefox 0.9.3 on 6 dual-processor SunFire V240's using "distcc" and GCC 3.3.1; using "gmake -j2" on just a single machine takes 58 minutes. While I suppose I was hoping for better than a 35% speedup by using 6 times as many hosts, compare & contrast this to using "qmake" from the Sun Grid Engine stuff - that inexplicably took an hour and 35 minutes (apparently it can only ship out top-level makes to the other CPUs - so most of the time, it's idle). So any noticable speedup is more than welcome.
On the other hand, I have to build Mozilla & Firefox with Sun's compilers for production use, because the Java plug-in for Netscape and Mozilla et al. is built with Sun's compilers, and it won't work if your Mozilla or Firefox is built with GCC. And, more importantly, our main in-house project (a giant image processing package) is built with the Sun compilers on Solaris (it's built with GCC - no, don't ask - on our Linux boxes, so I might get to try out "distcc" there), so if there's no way to get "distcc" going on Solaris with the Sun stuff, I may have to abandon the project for all but my own generic GCC-built toolsmithing.
Any hope Martin???
- Greg
__ distcc mailing list http://distcc.samba.org/
To unsubscribe or change options: http://lists.samba.org/mailman/listinfo/distcc
