Great job in front of the camera, actually. Never once did you strike me as nervous or unsure.extraketchup wrote:Forgive my awkwardness in front of the camera
extraketchup wrote:The labexec script... Here it is:
Oh, since I'm posting it, let's just go ahead and release it under GPL
.....
That's a good point, distcc operates at roughly 85%. You could probably eat up soem of the final 15% with this method, using some kind of program/script and a database of the already-compiled packages. It might also save on network usage.wescott wrote:extraketchup wrote:The labexec script... Here it is:
Oh, since I'm posting it, let's just go ahead and release it under GPL
.....
Hey Ketchup, you mentioned earlier that all your labs are identical. Instead of compiling the source (with distcc) 20 times (once per computer), wouldn't it be easier for 1 distcc compile of the source, quickpkg it, then emerge -K the provided custom binary package off the servers distfiles directory? If you already started doing this, than I guess there is more behind what you do with the script.

You might think of ogg as being difficult; but in circles such as these, it's not. We know what's best for us.playfool wrote:wow.. gwn caused a gig of additional transfer, that's amazing. My access log is quite entertaining - everybody went straight for the high quailty OGg Theora file, might I extend my thanks to all who supported the open and free formats.



Wow, I had no idea this thread was still active. What you suggest is exactly what I do (and have done all along). I compile on my (teacher) computer using the student workstations to speed up the process (tremendously). The /usr/portage/packages directory actually sits on NFS. I then "activate" an upgrade script (super-simple & home-made) which is run when the student workstations reboot, updating them using the latest packages on the server. The only downside is that this doesn't work so well outside of the lab, because our teacher computers are more of a mish-mash of hardware, so I am working on a separate "generic" package repository (i686 vs Athlon, and no fancy CFLAGS / USEFLAGS) for the non-athlon hardware we have. I definitely am 100% behind the compile once, distribute many times (bin packages) approach.Hey Ketchup, you mentioned earlier that all your labs are identical. Instead of compiling the source (with distcc) 20 times (once per computer), wouldn't it be easier for 1 distcc compile of the source, quickpkg it, then emerge -K the provided custom binary package off the servers distfiles directory? If you already started doing this, than I guess there is more behind what you do with the script.