rsync for cray t3e's
Sean Finney
finney at cs.swarthmore.edu
Sat Aug 4 12:26:18 EST 2001
On Sat, Aug 04, 2001 at 10:52:06AM +1000, Martin Pool spoketh:
> > Is there anyone who'd be interested in the diffs, if only for the sake
> > of porting it to another os?
>
> Yes, I would. Tridge just fixed build problems on a Cray SV ? (or J?)
> a while ago.
After taking another look at what was causing the problem, I rewrote the
changes so that there are only two files with minor changes: flist.c and
socket.c. I'll attach the diffs, let me know if you want me to do
anything else (like cvs-related or something).
>
> If it's possible under your site's policy, it might be nice to add
> that machine to the rsync build farm:
>
> http://build.samba.org/
I'll ask, though it's not likely, as I'm just a summer student who just
wanted to install rsync in his home directory :)
--sean
-------------- next part --------------
500a501
> u_long t3efix;
522a524,527
> /* t3efix:
> cc complains:
> CC-139 cc: ERROR File = socket.c, Line = 524
> Taking the address of a bit field is not allowed.
524a530,536
> ^
> so we just plop the data into some abitrary u_long and then
> re-assign it to ret.s_addr as a workaround
> */
>
> memcpy(&t3efix, hp->h_addr, hp->h_length);
> ret.s_addr=t3efix;
-------------- next part --------------
153c153,159
< if ((l = readlink(Path,Linkbuf,MAXPATHLEN-1)) == -1) {
---
> /* t3efix:
> readlink, according to its manpage, takes Path as a char*, whereas
> Path in this function is a const char*. quickfix is a (char*) cast,
> though I don't know if this would cause a problem later on (I don't
> think readlink modifies path)
> */
> if ((l = readlink((char*)Path,Linkbuf,MAXPATHLEN-1)) == -1) {
More information about the rsync
mailing list