>On 4/16/09, Leo Butler <l.butler at ed.ac.uk> wrote:
>
>> Hello, in the last few days since I first built 5.18.0 with
>> gcl-2.6.8pre, I have seen these `make_writable' errors. I believe they
>> are due to a bug in gcl. However, I do not see these errors with
>> 5.17.1
>> built with gcl-2.6.8pre.
>>
>> These errors also seem to be close to non-deterministic.
>>
>> Before I added the patch to mload.lisp suggested by Robert Dodier to
>> fix the
>> bug in `batch', this error did not happen during `make check'.
>
>I tried 5.18.1 + GCL 2.6.8 + patch for "batch" on Linux and I don't see
>the make_writable error in make check. Maybe someone else can try it.
>
>By the way, the GCL version is a little confusing. IIRC I pulled
>2_6_8pre or some tag like that from CVS, but it advertises itself
>in its banner as 2.6.8. Dunno what the story is there.
Robert,
I built 5.18.1 with gcl-2.6.8pre sans `batch' patch, and I had no
make_writable errors during `make check', too.
However, I have a script that behaves quite oddly with this maxima:
-When invoked in maxima via load("foo.mac"), the make_writable error appears
predictably after about 5 minutes.
-When the script is run from the command line via `maxima -b foo.mac' the
script runs fine.
(It has been running for about 18hours at the moment, and I expect it'll run for
10+days, barring further errors).
Leo
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.