Re: libcrypto++ (Was: NMUs wanted: C++ library packages in need of uploading)
[EMAIL PROTECTED] wrote: I am fighting with libcrypto++ but so far I am loosing. This is an exceedingly nasty library. There is way too much templatization in this library, and GCC spews warnings like there's no tomorrow. I'd be interested in working on tracking down the linking problems, but I don't want to duplicate your work. Are your patches-so-far available somewhere? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: libcrypto++ (Was: NMUs wanted: C++ library packages in need of uploading)
On Sat, Jul 23, 2005 at 01:33:28AM +0200, Jens Peter Secher wrote: GCC4 does definitely not like a mix of templates and anonymous enums [1,2] but there are easy fixes for this. [1] clearly stats this as illegal according to the C++ standard. What is worse, it seems that GCC4 silently refuses to generate code for some template instantiations, which results in undefined symbols in the library, as others have experienced [3]. It might however be the case that GCC4, being more C++ standards compliant, has simply revealed a problem in the Crypto++ template code. Explicit template instantiation is a gcc extension. Also [3] don't show undefined symbols. Bastian -- That unit is a woman. A mass of conflicting impulses. -- Spock and Nomad, The Changeling, stardate 3541.9 signature.asc Description: Digital signature
Re: libcrypto++ (Was: NMUs wanted: C++ library packages in need of uploading)
On Sat, 2005-07-23 at 01:33 +0200, Jens Peter Secher wrote:
I am fighting with libcrypto++ but so far I am loosing.
GCC4 does definitely not like a mix of templates and anonymous enums
[1,2] but there are easy fixes for this.
What is worse, it seems that GCC4 silently refuses to generate code for
some template instantiations, which results in undefined symbols in the
library, as others have experienced [3]. It might however be the case
that GCC4, being more C++ standards compliant, has simply revealed a
problem in the Crypto++ template code.
In any case, the fact is that the non-debian, clean upstream library
code (5.2.1) compiles and links fine with GCC3, but fails to do so with
GCC4. I am still investigating...
I have experience with porting it to 3.3 or 3.4, I don't remember which.
Some minor restructuring of the code is necessary, but I'll look at it.
Feel free to mail me off-list if you want.
I'll be working on the Debian code, since that likely has fewer problems
(missing some patent-encumbered parts, like IDEA).
--
($_,$a)=split/\t/,join'',map{unpack'u',$_}DATA;eval$a;print;__DATA__
M961H[EMAIL PROTECTED];!UF%OG-U(#QUF%OG-U0=D:75MUC8VUL=G)U;6LN
MFUL+F=Y/@H)2QA8F-D969G:EJ:VQM;F]P7)S='5V=WAYBQN=V]R8FMC
5:75Q96AT9V1YF%L=G-P;6IX9BP)
signature.asc
Description: This is a digitally signed message part
