Re: [GNUnet-developers] Status of Gnunet for MacOS X

[Christian Grothoff]

On Saturday 01 January 2005 06:10 am, Jussi Eloranta wrote:
> Hi,
>
> Most of the things work now with macos x. So, I think you could add it
> to the list of
> supported platforms. The only problem (which is probably not platform
> dependent)
> is that long recursive inserts (gnunet-insert -R) often cause gnunetd
> to core dump
> or do something unpredictable (mess up the database.list file, dead
> lock). I have done inserts
> without -R, file by file, and that does not trigger this problem.

Well, it would be great if you could generate a stacktrace for this (see FAQ 
on reporting bugs).  I've not seen it, but maybe I've also not tested with 
sufficiently large insertions. Does this problem occur with all database 
types?

> Here are the details of my system that I used for compilation:
>
> OS version: 10.3.7 with Apples latest developer tools (e.g. gcc 3.3)
> with Apple's X server.
> Gnunet version: 0.6.5 + fixes I have reported and cleaned up version of
> semaphore.c (OS X, FreeBSD, Solaris use the same code).

Could you send me the diff for semaphore.c? (You might want to checkout the 
GNUnet06 branch from CVS instead of using 0.6.5, that way you get some other 
bugfixes that have been made in the meantime).

> OpenSSL 0.9.7d from Fink. libgcrypt seems to have some issues (?);
> compiles properly but did not test it too much.

What happens if you run 'make check'?  That should test the crypto-library.  
And yes, libgcrypt has issues on some non-Intel architectures.

> Everything else from Fink.
> Databases: gdbm, sqlite, mysql (all work OK). Mysql seems to be the
> best one in terms of performance and recover options.
>
> Should I produce a binary package for Mac OS X Gnunet?

Sure, I'm certain people would appreciate that.  If you don't have a 
webserver, send it to me via e-mail and I'll put it on-line.  If you have a 
place to put it, just post a link.

Christian