[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gcl-devel] Make still fails on WinXP for ANSI

From: Camm Maguire
Subject: Re: [Gcl-devel] Make still fails on WinXP for ANSI
Date: Tue, 31 Aug 2010 11:41:12 -0400
User-agent: Gnus/5.11 (Gnus v5.11) Emacs/22.2 (gnu/linux)


Donald Winiecki <address@hidden> writes:

> Hi Camm and everyone,
> Apologies for the appearance of non-activity on my part.  My WinXP machine 
> has been struggling for awhile and
> has given up the ghost -- so to say.  I knew this was going to happen 
> eventually but didn't expect it
> presently.  (It's struggles may have been behind my inability to build a 
> GCL-ansi variant on WinXP.)

If there was a serious possibility of some systemic, e.g. memory
failure on this machine, I suggest we take this as a working
hypothesis given that vista goes through OK.  (arguing against this
however that your failures appear to have been in the same place on
multiple occasions.)

I think it is therefore far more productive to try to verify the app
builds on vista.  In particular, it would be really great if you could
work with Gaby on the open axiom bit, as the traditional axiom still
has a lot of non-lisp bsd dependent C code in it which makes it
impossible for me to verify under wine.  I have been able to verify
the maxima, acl2, and hol88 builds under wine.

The only real question with the win32 port is whether we go with the
new relocation file sfaslcoff.c or default to the older generic file
with all its foibles.  If we had a reproducible error, the next step
would be to swap out the loader code. 

> Since the only reason to try and get an WinXP machine up and running would be 
> to help out with WinXP tests of
> GCL -- and I don't know who might be using WinXP still -- I'm not sure it's 
> worth an entire machine to do
> this. 
> If you all think it is important to have an winXP-proven build I can pursue 
> putting a virtual machine on my
> Vista box and install XP into it -- but event this wouldn't be a `real' WinXP 
> installation.
> What do you all think?  Do we try to make sure it works on WinXP or do we 
> stop at 2.6.7 for XP?

In the mean time, I will ask my colleagues at ut if they have a win
box that might be used.

Take care,

> _don
> On Tue, Aug 24, 2010 at 10:44 AM, Camm Maguire <address@hidden> wrote:
>     Greetings!
>     Donald Winiecki <address@hidden> writes:
>     > Camm,
>     >
>     > I verified the source of CVS files to gcl268pre but make still fails 
> when building the ANSI variant. 
>     Here is
>     > a link to the make log.
>     >
>     > 
> https://sites.google.com/a/boisestate.edu/gcl-win32/home/build-logs/xp-ansi-m.log?attredirects=0&d=1
>     >
>     OK, this report is more what I expected.  To proceed further, I need
>     to guide you through some remote debugging steps.
>     make clean
>     ./configure --enable-debug --enable-ansi
>     make
>     cd pcl
>     echo '(load "sys-package.lisp")' '(setq *features* (delete (quote :kcl) 
> *features*))' '(load
>     "defsys.lisp")(push (quote :kcl) *features*)' '(setq 
> pcl::*default-pathname-extensions* (cons "lisp"
>     "o"))' '(setq pcl::*pathname-extensions* (cons "lisp" "o"))' '(load 
> "sys-proclaim.lisp")' '(setq
>     compiler::*default-h-file* t)' '(setq compiler::*default-c-file* t)' 
> '(setq compiler::*default-data-file*
>     t)' '(setq compiler::*default-system-p* t)' '(setq compiler::*keep-gaz* t 
> compiler::*tmp-dir* "")' '
>     (pcl::compile-pcl)' >foo
>     gdb ../unixport/saved_gcl.exe
>     (gdb) b call_init
>     (gdb) r
>     > (load "foo")
>     Then at the first break, please do
>     (gdb) display/x memory->cfd.cfd_start
>     and keep a list of which .o file is loaded at which start address.
>     Then
>     (gdb) c
>     to continue, and keep pressing 'c' until the error occurs.  gdb will
>     break and give you a fault address, please note this.  Then
>     (gdb) bt
>     to get a backtrace.
>     Please then leave this window open, send me the list of addresses, the
>     fault address, and the backtrace, and I can guide you further from
>     there.
>     Thanks again so much!
>     > _don
>     >
>     > _______________________________________________
>     > Gcl-devel mailing list
>     > address@hidden
>     > http://lists.gnu.org/mailman/listinfo/gcl-devel
>     --
>     Camm Maguire                                       address@hidden
>     ==========================================================================
>     "The earth is but one country, and mankind its citizens."  --  Baha'u'llah

Camm Maguire                                        address@hidden
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah

reply via email to

[Prev in Thread] Current Thread [Next in Thread]