|
|
Blitz Devel : |
From: Theodore Papadopoulo (Theodore.Papadopoulo_at_[hidden])
Date: 2003-01-13 07:53:09
cummings_at_[hidden] said:
> First, has the default template depth parameter in gcc changed
> recently?
2002-01-10 Jakub Jelinek <jakub_at_[hidden]>
* decl2.c (max_tinst_depth): Increase default limit to 500.
gcc-3.1 was released may, the 15th in 2002.
> But now it seems that this setting is lower than the current default
> value and is causing other codes to fail to compile with gcc 3.2.
I never said that it happened. I said that it might happen.
> Second, is the flag -fstrict-aliasing on by default with -O?
Not with -O with -O2.
That a more complicated story. This was true in 1998 when the flag
was introduced, then disabled because of some bugs and is now
re-enabled.
2002-03-20 Michael Meissner <meissner_at_[hidden]>
* doc/invoke.texi (Optimize Options): Document that -O2 sets
-fstrict-aliasing.
>From the date, this should be present in gcc-3.1. Now this is
somewhat close to the release. One might hope that the branch was not
yet forked at that time. But note also that this is just a
documentation update, so that the actual patch enabling the flag at
-O2 predates for quite some time the date of this patch. I'm sure
this is true for 3.1.
> Third, is the -fno-gcse flag still needed when compiling the sort of
> expression templates code that blitz uses?
It would be quite interesting to know for which reason it was
introduced in blitz... Efficiency, compilation problems ???
Theo.
--------------------------------------------------------------------
Theodore Papadopoulo
Email: Theodore.Papadopoulo_at_[hidden] Tel: (33) 04 92 38 76 01
--------------------------------------------------------------------