bug-cvs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

cvs output bug - missing 4096 byte blocks


From: Greg Klanderman
Subject: cvs output bug - missing 4096 byte blocks
Date: Fri, 14 Mar 2003 14:59:53 -0500

Hi,

I'm not absolutely certain whether this is a cvs bug or not but wanted
to post it here to see if anyone has any ideas or can reproduce it.

When running "cvs diff", if the diff is large enough, the output is
often missing 4096 byte blocks at byte offsets that are multiples of
4096.

* dual 2.8 Ghz P4 box with 4 Gb ram running RH 7.3.
* just downloaded and built cvs 1.11.5.
* using client cvs with :ext: method to a server on our local network.
* more prevalent when the machine is loaded.
* more prevalent with "-z3".

I only see the problem with CVS_RSH=ssh and when cvs' stderr is
combined with stdout (using "2>&1" in the shell, or running pcl-cvs
under emacs).

Here's an example; the first four runs are correct, after that things
go downhill:

phl| for i in {1..100} ; do
echo $i
/phl/build/cvs-1.11.5/src/cvs -f -z3 diff -u taxes-intl.qcf 2>&1 | tee 
cvs.out.$i | wc
done
1
   5322   25355  207275
2
   5322   25355  207275
3
   5322   25355  207275
4
   5322   25355  207275
5
   5206   24847  203133
6
   5216   24852  203122
7
   2907   13739  112390
8
   5120   24343  199063
9
   4801   22830  186672
10
   1732    8232   67155

The same problem is occurring for "cvs annotate" and I suspect other
commands too (update, etc):

phl| for i in {1..100} ; do
echo $i
/phl/build/cvs-1.11.5/src/cvs -f -z3 annotate taxes-intl.qcf 2>&1 | tee 
cvs.out.$i | wc   
done
1
  25073  199469 1880129
2
   4660   36702  340728
3
  24355  194153 1826345
4
  15590  123161 1156043

thanks for any assistance,
greg




reply via email to

[Prev in Thread] Current Thread [Next in Thread]