Re: freeze for sbcl-2.6.3

Andreas, I think you do have a valid point. A ppc64 configuration exists
that crashes on startup. The problem I have is in the approach you took.
First of all you could have started with a question like "is ppc64 known to
be building fine?" rather than just "fix ppc64" when it ostensibly isn't
broken.

So not having started with the question, you got right into a diagnosis
which is almost 100% wrong. Claiming x86-64 as different is wrong, and even
if were an exception, that doesn't explain all the other platforms that
work.  And anything about elfcore is flat-out wrong because nothing uses
elfcore by default. And no platform avoids calling through the linkage
table.  So when we are presented with all these dubious "facts" it casts
doubt on everything else you're saying. Finally, the patch itself isn't
what I would have done: if foreign-reinit needs to be called sooner, just
call it sooner.

A way to have expressed yourself so that developers don't dismiss it as
AI-slop would have been more like:
- I think there is a configuration of ppc64 which can crash on startup
because <blah>.  Granted this is an unusual configuration, one which has no
controlling TTY.  When that occurs, then sqrt gets called before it has a
linkage table entry through the following chain of calls

 And then maybe ask how to fix it instead of just sending "fix ppc64". I
think your approach caused >1 developer to just disregard that patch if not
also the other well-intended patches

_______________________________________________
Sbcl-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sbcl-devel