On Tue, 28 Aug 2001 12:13:40 -0400, you wrote:
>> /*** anyway, pthread_once is not too good either - something
>> like a C++ constructor on a global static variable would be
>> much better ***/
>> pthread_once(&tid_once, init_openssl_tid);
>
>That's not portable -- go look at the netscape/mozilla guidelines.
>Particularly if you'ure putting openssl in a shared library. And
>requiring C++ for threads to work seems like a very bad thing.
Too bad there's no standard on what dlopen() should do with
"constructors" sections. But we can always ./configure anything we want,
can't we? Oh well...
>> /* this is not 100% kosher, but still ... */
>> return (unsigned long) pthread_getspecific(tid_key);
>
>C'mon. You can't complain about pthread_t/ulong and not feel bad trying
>to do void*/ulong.
Believe me, I do feel bad about it :-) At least, most compilers won't
choke on it even if pthread_t is not scalar. Well, scrap it anyway.
> At the cost of one ulong-sized indirect memory
>fetch, the code I wrote is completely correct and portable.
Absolutely.
> /r$
Cheers!
Dima
______________________________________________________________________
OpenSSL Project http://www.openssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
