update to tested platform list
Morgan Burke
SGI running IRIX 6.5.3 (1.11.1)
Although the cvs binary appears to run correctly (haven't used it for
long), there are several bugs in the make procedure --
Both "configure" and "make" terminate with the same error message:
Cannot access +x: No such file or directory
Invalid mode
which appears to be caused by the command
chmod -f +x
(-f is not a recognized switch to chmod in Irix). While trying to
figure out if these errors were fatal or not, I ran the tests
("make check"), which hung right after the lines:
Making check in src
make check-local
/bin/sh ./sanity.sh `pwd`/cvs
This test should produce no other output than this line, and a final "OK".
<<< HUNG HERE WITH NO "OK"
But as I said, the binary appears to work for my purposes (client/server
checkouts and commits), and it fixes a problem I was having with v1.9.6,
so I guess I'm a satisfied customer.
-- Morgan Burke
mor...@sitka.triumf.ca
PS. the Irix 6.5 pre-compiled binary on cvshome.org ought to be compiled
for the R4000 (not R12000) to be executable on the widest range of SGI
systems.
Larry Jones
>
> Both "configure" and "make" terminate with the same error message:
>
> Cannot access +x: No such file or directory
> Invalid mode
Already fixed in the current development version. (The fix is, as you
suggest, to remove the -f from the chmod commands.)
> Making check in src
> make check-local
> /bin/sh ./sanity.sh `pwd`/cvs
> This test should produce no other output than this line, and a final "OK".
> <<< HUNG HERE WITH NO "OK"
Are you sure it hung? The tests take a *long* time to run (there are a
lot of them), you may not have waited long enough.
> PS. the Irix 6.5 pre-compiled binary on cvshome.org ought to be compiled
> for the R4000 (not R12000) to be executable on the widest range of SGI
> systems.
A good point. I'll try to remember that in the future.
-Larry Jones
It's like SOMEthing... I just can't think of it. -- Calvin
Derek R. Price
> > Making check in src
> > make check-local
> > /bin/sh ./sanity.sh `pwd`/cvs
> > This test should produce no other output than this line, and a final "OK".
> > <<< HUNG HERE WITH NO "OK"
>
> Are you sure it hung? The tests take a *long* time to run (there are a
> lot of them), you may not have waited long enough.
'tail -f src/check.log' can give you some idea of what's happening. Be a
little careful though, 'make check' produces two check.logs, so about halfway
through the tests you have to abort the first tail -f and run a second to keep
getting status.
'top' or 'ps' might be helpful too, of course.
Derek
--
Derek Price CVS Solutions Architect ( http://CVSHome.org )
mailto:dpr...@collab.net CollabNet ( http://collab.net )
--
I will not conduct my own fire drills.
I will not conduct my own fire drills.
I will not conduct my own fire drills...
- Bart Simpson on chalkboard, _The Simpsons_
Morgan Burke
> > > This test should produce no other output than this line, and a final "OK".
> > > <<< HUNG HERE WITH NO "OK"
> >
> > Are you sure it hung? The tests take a *long* time to run (there are a
> > lot of them), you may not have waited long enough.
>
> 'tail -f src/check.log' can give you some idea of what's happening. Be a
> little careful though, 'make check' produces two check.logs, so about halfway
> through the tests you have to abort the first tail -f and run a second to keep
> getting status.
The check.log definitely seems to come to a stop after about 20 minutes
of crunching. (Tail appended to this message.) Restarting tail -f gives
me nothing new, and top shows no activity.
If I put "make check" into the background, then it stops at the same point
but shell job control pops up a message "suspended on tty input". I brought
it back to the foreground and tried a few generic inputs (y, n, CR) but
got no response.
-- Morgan Burke
mor...@sitka.triumf.ca
TAIL OF check.log:
PASS: ann-id-5
Annotations for m
***************
1.2 (morgan 04-Jun-01): $Id: m,v 1.1 2001/06/04 19:56:10 morgan Exp $
1.2 (morgan 04-Jun-01): line2
PASS: ann-id-6
cvs [checkout aborted]: /tmp/cvs-sanity/crerepos/CVSROOT: No such file or directory
PASS: crerepos-1
cvs [checkout aborted]: /tmp/cvs-sanity/crerepos/CVSROOT: No such file or directory
PASS: crerepos-2
cvs checkout: cannot find module `cvs-sanity' - ignored
PASS: crerepos-3
PASS: crerepos-4
You have [0] altered files in this repository.
Are you sure you want to release (and delete) directory `CVSROOT': PASS: crerepos-5
PASS: crerepos-6
(plus one blank line - mb)
Larry Jones
>
> The check.log definitely seems to come to a stop after about 20 minutes
> of crunching. (Tail appended to this message.) Restarting tail -f gives
> me nothing new, and top shows no activity.
>
> If I put "make check" into the background, then it stops at the same point
> but shell job control pops up a message "suspended on tty input". I brought
> it back to the foreground and tried a few generic inputs (y, n, CR) but
> got no response.
Very interesting.
> TAIL OF check.log:
[...]
> PASS: crerepos-6
You're hanging on crerepos-6a, which is a very strange test. It tries
to a checkout with CVSROOT set to ../crerepos, which CVS interprets
(strangely enough) as :ext:..:/crerepos and thus ends up doing:
rsh .. cvs server
even though it's a *local* test, not a remote one. This apparently
prompts any number of different error messages from various rsh
implementations, but on yours it apparently does *not* produce any error
message, but just quietly starts reading from stdin instead. What
happens if you type "rsh .."?
I'm thinking that for this test we should probably set CVS_RSH to
something innocuous like "false". In the meantime, feel free to comment
out that particular test so that you can finish running the test suite.
-Larry Jones
The surgeon general should issue a warning about playing with girls. -- Calvin
Morgan Burke
> > If I put "make check" into the background, then it stops at the same point
> > but shell job control pops up a message "suspended on tty input". I brought
> > it back to the foreground and tried a few generic inputs (y, n, CR) but
> > got no response.
>
> Very interesting.
>
> > TAIL OF check.log:
> [...]
> > PASS: crerepos-6
>
> You're hanging on crerepos-6a, which is a very strange test. It tries
> to a checkout with CVSROOT set to ../crerepos, which CVS interprets
> (strangely enough) as :ext:..:/crerepos and thus ends up doing:
>
> rsh .. cvs server
>
> even though it's a *local* test, not a remote one. This apparently
> prompts any number of different error messages from various rsh
> implementations, but on yours it apparently does *not* produce any error
> message, but just quietly starts reading from stdin instead. What
> happens if you type "rsh .."?
sitka>rsh ..
The authenticity of host '.. (0.0.0.0)' can't be established.
DSA key fingerprint is 43:95:a3:04:08:5b:6d:f0:a5:ea:98:84:9c:30:8a:96.
Are you sure you want to continue connecting (yes/no)?
sitka>ls -l `which rsh`
lrwxr-xr-x 1 root sys 3 Mar 16 2000 /usr/local/bin/rsh -> ssh
heh heh... A little security "patch" from our sysadmin.
When I fix it to use the real rsh, the tests complete OK.
I appreciate your attention on this. FYI, I've been using CVS 1.11.1p1
in spite of all this, doing cross-continent remote updates and
whatnot without any problems.
Thanks,
-- Morgan Burke
mor...@sitka.triumf.ca
Derek R. Price
> I'm thinking that for this test we should probably set CVS_RSH to
> something innocuous like "false". In the meantime, feel free to comment
> out that particular test so that you can finish running the test suite.
I just checked this in. Since this is a local test and not a remote, I'm
assuming we made an implicit promise not to call RSH.
Derek
--
Derek Price CVS Solutions Architect ( http://CVSHome.org )
mailto:dpr...@collab.net CollabNet ( http://collab.net )
--
Southern DOS: Y'all reckon? (yep/Nope)
Larry Jones
>
> sitka>rsh ..
> The authenticity of host '.. (0.0.0.0)' can't be established.
> DSA key fingerprint is 43:95:a3:04:08:5b:6d:f0:a5:ea:98:84:9c:30:8a:96.
> Are you sure you want to continue connecting (yes/no)?
Which would explain why the test hung -- it was waiting for you to
answer the question.
> sitka>ls -l `which rsh`
> lrwxr-xr-x 1 root sys 3 Mar 16 2000 /usr/local/bin/rsh -> ssh
Interesting. Real rsh generally complains that .. isn't a know host;
ssh seems to interpret it as 0.0.0.0, which in turn in generally
interpreted as 127.0.0.1 (i.e., localhost).
-Larry Jones
Everything's gotta have rules, rules, rules! -- Calvin
Martin Neitzel
MB>
MB> PS. the Irix 6.5 pre-compiled binary on cvshome.org ought to be compiled
MB> for the R4000 (not R12000) to be executable on the widest range of SGI
MB> systems.
I am building bleeding edge cvs binaries about every two months or
so on my lowly 32bit R3000 SGI with IRIX 5.3. People call me crazy
for supporting this platform but I was pretty successful so far with
preparing binaries for today's high-end 6.5.x SGIs on my box.
If anybody wants cvs binary kits, I could learn how to prepare proper
"inst" packages. The only caveat I have right now is that I'll have yet
to install a current automake, and for that some perl-5, and for that...
well, you get the picture. Give me three weeks and I'll have them kits
for anybody who's interested.
Martin
Derek R. Price
> Morgan Burke writes:
> > sitka>ls -l `which rsh`
> > lrwxr-xr-x 1 root sys 3 Mar 16 2000 /usr/local/bin/rsh -> ssh
>
> Interesting. Real rsh generally complains that .. isn't a know host;
> ssh seems to interpret it as 0.0.0.0, which in turn in generally
> interpreted as 127.0.0.1 (i.e., localhost).
That could be a function of Morgan's particular ssh implementation/version -
my ssh complains properly about 'ssh ..':
[dprice@empress doc]$ ssh ..
ssh: ..: Name or service not known
[dprice@empress doc]$
Derek
--
Derek Price CVS Solutions Architect ( http://CVSHome.org )
mailto:dpr...@collab.net CollabNet ( http://collab.net )
--
I wouldn't bring up Paris right now. It's poor salesmanship.
- Humphrey Bogart as Rick, _Casablanca_