On Thu, 1 Feb 2018 10:21:44 +0000 James Cowgill <jcowg...@debian.org> wrote: > Hi, > > On 27/01/18 19:13, Gianfranco Costamagna wrote: > > On Fri, 5 Jan 2018 16:12:14 +0100 Gianfranco Costamagna > > <locutusofb...@debian.org> wrote: > >> On Thu, 4 Jan 2018 14:30:08 +0000 James Cowgill <jcowg...@debian.org> > >> wrote: > >>> Hi, > >>> > >>> On 04/01/18 14:26, Gianfranco Costamagna wrote: > >>>> Hello, > >>>> > >>>>> Is there a way for me to run your testcase without ghc-stage1 (which > >>>>> would require building ghc)? > >>>> > >>>> I don't know, however I attached ghc-stage1 binaries to the report > >>> > >>> The "ghc-stage1" you attached is a shell script which execs some other > >>> binary which you didn't attach. > >>> > >> > >> this is true, I hope you can get from my home in eller.debian.org :) > >> > > > > hello, ping? > > (this is the last blocker for ghc 8.2 in unstable, unfortunately somebody > > would like to remove from mips instead of delaying the transition any > > longer) > > Sorry I got a bit sidetracked working on other things. > > I have a slightly reduced testcase, but it still requires about 50M of > objects so I'm working on reducing it even more.
Cool, thanks. > I did a test compile of ghc using the bfd linker and it does build (pass > --disable-ld-override to ghc configure). This might be usable as a > workaround if you don't want to delay the transition. I expect this will > make ghc / haskell build times increase a lot on mips* though. Did you try that on mips64el? Gianfranco tried on mips or mipsel IIRC and it was running out of memory. Perhaps reducing the optimization levels or the debugging symbols could help... Emilio
