When running emerge pine, and attempting to build Pine 4.60, the build fails due to lack of RAM. It requires approxamately 512MB (or more) to build. This bug has been acknowledged at the UW website and they have released a fix which should probably be put into the portage. Users without lots of RAM will more likely then not experience this often, as gcc appears to not like to touch the swap space. Reproducible: Always Steps to Reproduce: 1. Boot Linux with 128 MB of RAM (if you have more, use mem=128M) 2. 'emerge pine' 3. Watch it fail on filter.c. Actual Results: When gcc encounters filter.c, it's memory usage will climb continously until it reaches an amount where all free RAM is used. In the process of requiring more memory, the Linux kernel will swap applications to swap-space, freeing as much as it can (down to about 10MB short of all system RAM, in this case). After it fails to fulfill the RAM requirements of gcc using this method, it starts killing things it can't swap to the swap space. When this fails, it kills the apparently rogue gcc. This is a bug with filter.c - see PINE website for information on that. Expected Results: Compiled successfully. The appropriate fix in this case is to patch filter.c (which I have not the experience to do) or to upgrade to PINE 4.61 in the portage tree. Information from 'emerge info' not relevant. System memory is not faulty, was tested for 48+ hours using memtest86. System has a 550MHz PIII and 128MB of RAM with 1 GB of swap space.
4.61 aready in portage.
Should we pull 4.60 from portage?
not until we have these keyword in 4.61: ppc x86 sparc amd64. Andrej, if you could, please file bugs assign to those arches asking for stable keywords.
I can make x86 stable, I will file a bug for the rest.
Ok, 4.61-r2 stabilized for all arches 4.60 is stable for, so I'm pulling everything below 4.61-r2. I also did a cleanup in files/ dir by removing no longer used patches. Thanks for reporting, Michael.
No problem, andrej. I failed to notice the slew of e-mail generated by this bug because I haven't checked this inbox for a few days, and probably should've piped up when I found out that 4.61 worked as expected on x86, since that's what I've been using for a while now :-D. Yay for die-hard PINE fans! It's rather unfortunate that the other bugs on this issue failed to pinpoint WHY this behavior was exhibited, and I haven't seen or checked (yet) to see if those were marked resolved, but they probably should be now since 4.61 was marked stable. Good day.