Simon Peyton-Jones (simonpj@xxxxxxxxxxxxx) wrote:
> I don't know whether you did as I suggested, and checked out a
> completely fresh tree? I strongly urge you to do so. Doing an update
> into an old tree is vulnerable to (a) new directories not getting
> checked out and (b) old files from the previous build messing up the new
> build. The make system can't reasonably deal with (b).
Yes, I deleted my old folder, and checkout fresh from the CVS:
cvs -ro ghc-6-2-branch fpconfig
cd fptools
cvs -ro ghc-6-2-branch ghc hslibs libraries
Is this OK?
I also checkout happy-1.14 and have --enable-srs-tree-happy (happy is found and
built nicely).
> (b) old files from the previous build messing up the new
> build. The make system can't reasonably deal with (b).
>
What to do when ggc segfaults in the middle of the process?
a) start from the scratch or
b) just continue with the "make"?
I'm doing b) taking care that all temporary files are deleted.
> I think it'd save you time to start from a fresh tree. If that is what
> you have done, I'm much more puzzled -- all this stuff works here.
I'm also puzzled since I already (in the past) passed this stage and the process
was dying at stage2. Now, I even cannot reproduce the old state :-(
I'm doing build with the SplitObjs = NO entry in build.mk. Is it OK?
All in all, it seems (to me) that until know I'm just hitting some non-MSYS
issues in trying to build ghc. but I won't give it up.
I am waiting for your reply and then I'll try again with the fresh tree.
Sincerely,
Gour
--
Gour
gour@xxxxxxxxxxxx
Registered Linux User #278493
|
