Robert Dodier <robert.dodier at gmail.com> writes:
> On 2013-03-18, Rupert Swarbrick <rswarbrick at gmail.com> wrote:
>> Maybe I didn't make myself clear, but I think that the current state of
>> affairs is considerably *worse* on GCL than what I'm proposing.
>
> Well, my message was probably too strongly worded. But anyway if there
> is a way to get the right stuff built, I hope we'll do that. Beyond that
> I won't worry about it too much.
The problem is that, in order to make sys-proclaim.lisp you need to load
up all of Maxima (without the old sys-proclaim.lisp, since it might be
wrong). Then, to make use of it, you need to recompile the whole lot
with the new sys-proclaim.lisp.
As such, I *think* that the only way to ensure that the right stuff is
built every time on GCL, when .lisp files can change "randomly" is to
build everything twice. Yuck.
Since I'm guessing that no-one's keen on that solution, I'm going to
carry on advocating building sys-proclaim.lisp just once in each tree!
:-)
Thinking about it, for a bit of extra safety maybe we could have a git
hook that deleted sys-proclaim.lisp in the user's build tree if a lisp
file was changed on a checkout. I'm not sure how keen people would be on
this though.
Rupert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 315 bytes
Desc: not available
URL: <http://www.math.utexas.edu/pipermail/maxima/attachments/20130319/b08f3917/attachment.pgp>;