Re: libpq not reentrant

Il ven, 2002-01-18 alle 16:35, Tom Lane ha scritto:
> Federico Di Gregorio <fog(at)initd(dot)org> writes:
> >> Given that as of 7.2, MD5 is the preferred password encryption method
> >> and crypt() is deprecated, I'm not inclined to spend a lot of work
> >> trying to develop a bulletproof autoconf procedure for making crypt
> >> re-entrant. I'm strongly inclined to just document the problem and
> >> leave it at that. Comments?
>
> > yes. what about using crypt from DES (reentrant) if postgres is being
> > linked with it?
>
> Dunno, how would we detect that? Is it any more portable than crypt_r?

./cofigure --help
...
--with-openssl[=DIR] build with OpenSSL support [/usr/local/ssl]

so the right library is already detected. the function is called
des_crypt and is completely compatible with std. i think that a

AC_SEARCH_LIBS(crypto, des_crypt)

in configure.in will suffice. then:

#ifdef HAVE_DES_CRYPT
#define crypt des_crypt
#endif

in the right file will do the trick. (i know this is not a real patch,
sorry. i can make one later at home if you need.)

> Is it worth the trouble?

i don't know if other drivers/applications have an aggressive
multithreaded approach as psycopg has, but if they do, yes, it is worth,
imo. if psycopg is the only one, no, because i will fix the problem in
it.

> > is the MD5 stuff reentrant?
>
> Yes. It's all our own code, so we don't have to rely on the vagaries of
> local libraries, either.

great.

--
Federico Di Gregorio
Debian GNU/Linux Developer & Italian Press Contact fog(at)debian(dot)org
INIT.D Developer fog(at)initd(dot)org
All'inizio ho scritto un programma proprietario, in esclusiva per il
cliente; è stato tristissimo, perchè mi ha succhiato un pezzo di
anima. -- Alessandro Rubini