2009/8/4 Simon King <simon.k...@nuigalway.ie>:
>
> Your implication that ECL should "note the error and then continue to
> compile" somehow reminds a recent thread (probably on sage-devel)
> where it was reported that some spkg sends compiler warnings to /dev/
> null.
Yes it was in sage-devel. I raised this as an objection to including
sandpiples - or rather the particular the library that sandpipes would
need, as it thinsg like
gcc foobar.c > /dev/null 2>&1
I'm also in the process of writing an imporove spkg-install for
'lcalc' as that was another package which tried to hide the warnings,
by passing a flag to the assembler to not display the warnings. It
failed to do this on the Sun assembler, as the Sun assembler does not
take such a flag. Instead it reported it did not understand the flag.
I seriously question why anyone would write code in such a way. I'm
not saying there are odd cases where code does things that a compiler
does not like, and it is the best eway to do it. But very often it is
just a lack of attention to detail, which I find odd given a lot of
the code in sage is written by Mathematicains, who I believe are
generally a group of professionals that pay attention to detail more
than most other professionals.
Since I'd enabled the warnings in lcalc, there are tons of reports of
unused variables, in additon to other warnings.
IMHO, it woulod be better if web browsers refused to display dodgy
HTML rather than try to do the best they can with it. Then it would
force web designers to write propor code, and not junk. (It would mean
the Mathematica page would not display at all, as that has numerous
warnings and errors).
> Unless you can *prove* that the special circumstance ("static typing
> error in which a program which, in some circumstance, would do a list
> operation on an argument which is an integer") can not occur, one
> must assume that it is indeed a bug (not of ECL).
I also think it sensible to wirte a comment in code if it is going to
generate a warning, and it is for a good reason.
I'm well aware compilers have improved over the years (one might argue
the converse with gcc), but some of these warnings should have been
spotted long ago.
> It is an impressive achievement of ECL to finally find it, after 35
> years!
Yes it is impressive.
I often wonder if there should be a 'feature freeze' in Sage and
devote some time to sorting out why these warnings are occuraing and
whether they could result in the generation of incorrect results.
Dave
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
