Just a FYI:

bash-2.05b$ gcc -Wall -o upgrader_0.6.exe ./193910-upgrader_0.6.c
193910-upgrader_0.6.c: In function `main':
193910-upgrader_0.6.c:466: warning: long unsigned int format, different type arg (arg 4)

A current cygwin install using:
bash-2.05b$ gcc --version
gcc (GCC) 3.3.1 (cygming special)

--Nathan

And here's the binary.

--Nathan

Ah.. cygwin must be using unsigned int instead of the more correct (?) unsigned long type for "off_t" (part of the struct returned by stat(2)).

I'll add a cast to the printf to eliminate the warning for v0.7 (someday).

Thanks

Did you ever post the upgrader as ported to MacOS (X, I presume)? Inquiring minds ...

Ahh, I see it now - I missed the "Attachment" link tucked away on the left.

OK - trying updater now on MacOS X (with Hijack, gunzip and ncftpput at the ready). When I try to apply a certain 3.0 alpha I get this:

Error, size mismatch: 3334488576 vs 0

Clues, anyone?

Recompiled it. For some reason fstat() wasn't returning anything useful. Working now.

You probably had the wrong binary, the first one I posted was useless

No problem at all. Looks like all is well now.

I wasn't getting any playlists in 3.0a5, and going back to 3.0a3 wasn't helping 'cause I still couldn't make it into jEmplode to do the rebuild. I went back to 2.0, then forward to 3.0a5 and the playlists were there <!>.

Of course now, hindsight's 20/15, so I'm running the command line rebuild database in jemplode v51 (which is probably what I should have run in the first place <LOL>).

Oh. Hm. I wonder if 3.0a5 can deal with the old (2.0) databases directly.

I guess I should steal a player binary off my spare player.

Yeah, it seemed odd to me, actually. But that's how it went down. Meanwhile, I'm back on 2.0 after all - I've got some more CDs to upload. Holiday backlog ... yule gotta love it <groan>.