* Blitz, Boost, FTensor failures with new C++ parser
@ 2003-01-08 22:16 Janis Johnson
2003-01-08 22:41 ` Gabriel Dos Reis
2003-01-09 13:40 ` Theodore Papadopoulo
0 siblings, 2 replies; 6+ messages in thread
From: Janis Johnson @ 2003-01-08 22:16 UTC (permalink / raw)
To: gcc
I've been building and testing Blitz, Boost, and FTensor occasionally
with the mainline on powerpc64-linux. Yesterday Blitz and FTensor
failed to build and Boost had new test errors, and I just confirmed that
they have the same problems today on i686-linux. Blitz, at least, gets
a lot of warnings about using deprecated features. These should all
continue to work with the new parser, shouldn't they, particularly the
ones that have been release criteria for GCC 3.x?
Janis
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Blitz, Boost, FTensor failures with new C++ parser
2003-01-08 22:16 Blitz, Boost, FTensor failures with new C++ parser Janis Johnson
@ 2003-01-08 22:41 ` Gabriel Dos Reis
2003-01-10 2:24 ` Janis Johnson
2003-01-09 13:40 ` Theodore Papadopoulo
1 sibling, 1 reply; 6+ messages in thread
From: Gabriel Dos Reis @ 2003-01-08 22:41 UTC (permalink / raw)
To: Janis Johnson; +Cc: gcc
Janis Johnson <janis187@us.ibm.com> writes:
| I've been building and testing Blitz, Boost, and FTensor occasionally
| with the mainline on powerpc64-linux. Yesterday Blitz and FTensor
| failed to build and Boost had new test errors, and I just confirmed that
| they have the same problems today on i686-linux. Blitz, at least, gets
| a lot of warnings about using deprecated features. These should all
| continue to work with the new parser, shouldn't they, particularly the
| ones that have been release criteria for GCC 3.x?
Some features have been depecreated in previous releases and nominated
for removal in future releases.
It would be interesting to see the kind of deprecated features that have
been removed and are causing you troubles. Implicit typename for
example have proven to make us generated *wrong code for valid
programs*, which questions the feature in the first place.
Boost is an actively developed library, the authors of which track
standard conformance.
-- Gaby
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Blitz, Boost, FTensor failures with new C++ parser
2003-01-08 22:16 Blitz, Boost, FTensor failures with new C++ parser Janis Johnson
2003-01-08 22:41 ` Gabriel Dos Reis
@ 2003-01-09 13:40 ` Theodore Papadopoulo
2003-01-09 21:14 ` Janis Johnson
1 sibling, 1 reply; 6+ messages in thread
From: Theodore Papadopoulo @ 2003-01-09 13:40 UTC (permalink / raw)
To: Janis Johnson; +Cc: gcc
janis187@us.ibm.com said:
> I've been building and testing Blitz, Boost, and FTensor occasionally
> with the mainline on powerpc64-linux. Yesterday Blitz and FTensor
> failed to build and Boost had new test errors, and I just confirmed
> that they have the same problems today on i686-linux. Blitz, at
> least, gets a lot of warnings about using deprecated features. These
> should all continue to work with the new parser, shouldn't they,
> particularly the ones that have been release criteria for GCC 3.x?
Take the cvs version of blitz from sourceforge, unless I made some
mistakes, it should compile perfectly. Actually, the bug related to
enum in default parameters that I submitted two or three days ago is
originated from blitz. The remaining problem I encountered is that at
some point blitz was making use of the following syntax:
template <typename T>
void f()
{
return T::type(...);
}
That does not work any longer. I did not report it because I believe
it is indeed wrong syntax (The last instance of the implicit typename
extension ?). I corrected this, so now everything should work just
fine.
In any case, if you find any problems with that version of blitz, let
me now, I'm interested (with a few others) in keeping blitz in
working state (and even improving it).
I wish my own code will be as easy to convert to 3.4.
Theo.
--------------------------------------------------------------------
Theodore Papadopoulo
Email: Theodore.Papadopoulo@sophia.inria.fr Tel: (33) 04 92 38 76 01
--------------------------------------------------------------------
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Blitz, Boost, FTensor failures with new C++ parser
2003-01-09 13:40 ` Theodore Papadopoulo
@ 2003-01-09 21:14 ` Janis Johnson
0 siblings, 0 replies; 6+ messages in thread
From: Janis Johnson @ 2003-01-09 21:14 UTC (permalink / raw)
To: Theodore Papadopoulo; +Cc: Janis Johnson, gcc
On Thu, Jan 09, 2003 at 10:51:21AM +0100, Theodore Papadopoulo wrote:
>
> janis187@us.ibm.com said:
> > I've been building and testing Blitz, Boost, and FTensor occasionally
> > with the mainline on powerpc64-linux. Yesterday Blitz and FTensor
> > failed to build and Boost had new test errors, and I just confirmed
> > that they have the same problems today on i686-linux. Blitz, at
> > least, gets a lot of warnings about using deprecated features. These
> > should all continue to work with the new parser, shouldn't they,
> > particularly the ones that have been release criteria for GCC 3.x?
>
> Take the cvs version of blitz from sourceforge, unless I made some
> mistakes, it should compile perfectly.
Thanks, I tried it with the mainline on i686-pc-linux-gnu and
powerpc64-unknown-linux-gnu and all of the tests succeeded. I'll
contact the Blitz development list about a couple of nits and to
ask if they can provide a snapshot for us to download.
Janis
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Blitz, Boost, FTensor failures with new C++ parser
2003-01-08 22:41 ` Gabriel Dos Reis
@ 2003-01-10 2:24 ` Janis Johnson
0 siblings, 0 replies; 6+ messages in thread
From: Janis Johnson @ 2003-01-10 2:24 UTC (permalink / raw)
To: Gabriel Dos Reis; +Cc: Janis Johnson, gcc
On Wed, Jan 08, 2003 at 11:15:26PM +0100, Gabriel Dos Reis wrote:
> Janis Johnson <janis187@us.ibm.com> writes:
>
> | I've been building and testing Blitz, Boost, and FTensor occasionally
> | with the mainline on powerpc64-linux. Yesterday Blitz and FTensor
> | failed to build and Boost had new test errors, and I just confirmed that
> | they have the same problems today on i686-linux. Blitz, at least, gets
> | a lot of warnings about using deprecated features. These should all
> | continue to work with the new parser, shouldn't they, particularly the
> | ones that have been release criteria for GCC 3.x?
>
> Some features have been depecreated in previous releases and nominated
> for removal in future releases.
>
> It would be interesting to see the kind of deprecated features that have
> been removed and are causing you troubles. Implicit typename for
> example have proven to make us generated *wrong code for valid
> programs*, which questions the feature in the first place.
Information like that ought to be included in the 3.4 release notes,
starting now, to help people updating their C++ code to use with 3.4.
> Boost is an actively developed library, the authors of which track
> standard conformance.
Boost builds but now gets some test failures.
Qt 2.3.0 builds, but I killed moc (an executable it builds) after it
had been running for a couple of hours. I haven't tried POOMA.
I don't know C++ well enough to track these things down and figure out
whether the failures are due to invalid code that the compiler used to
handle, or new bugs in the compiler.
Janis
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Blitz, Boost, FTensor failures with new C++ parser
@ 2003-01-10 0:15 Walter Landry
0 siblings, 0 replies; 6+ messages in thread
From: Walter Landry @ 2003-01-10 0:15 UTC (permalink / raw)
To: gcc
Janis Johnson wrote:
> I've been building and testing Blitz, Boost, and FTensor
> occasionally with the mainline on powerpc64-linux. Yesterday Blitz
> and FTensor failed to build and Boost had new test errors, and I
> just confirmed that they have the same problems today on i686-linux.
> Blitz, at least, gets a lot of warnings about using deprecated
> features. These should all continue to work with the new parser,
> shouldn't they, particularly the ones that have been release
> criteria for GCC 3.x?
If these problems are caused by implicit typenames, I have a new
version that should solve those problems:
http://www.oonumerics.org/FTensor/FTensor-1.1pre20.tar.gz
Regards,
Walter Landry
wlandry@ucsd.edu
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2003-01-10 0:41 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-01-08 22:16 Blitz, Boost, FTensor failures with new C++ parser Janis Johnson
2003-01-08 22:41 ` Gabriel Dos Reis
2003-01-10 2:24 ` Janis Johnson
2003-01-09 13:40 ` Theodore Papadopoulo
2003-01-09 21:14 ` Janis Johnson
2003-01-10 0:15 Walter Landry
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).