[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Difference in output of "cvs up" and "cvs up -p"
From: |
Raye Raskin |
Subject: |
Difference in output of "cvs up" and "cvs up -p" |
Date: |
Wed, 11 Oct 2006 14:35:09 -0700 |
Submitter-Id: net
Originator: rraskin@vicor.com
Organization: net
Confidential: no
Synopsis: "cvs up -p file > foo" adds an EOF to foo, "cvs up file" does not
add an EOF to file
Severity: non-critical
Priority: medium
Category: cvs
Class: sw-bug
Release: 1.11.17
Environment:
System: Linux sprocket.vicor.com 2.6.9-42.ELsmp #1 SMP Wed Jul 12 23:27:17 EDT
2006 i686 i686 i386 GNU/Linux
Architecture: i686
Description:
We have a .doc file that probably should have been -kb, but it wasn't.
On checkout on Linux the file is fine and there is no corruption. A
"cvs checkout" or "cvs update" of this file puts a file on the disk that
is 100% correct. However, a "cvs up -p" of this file, redirected to a
file, inserts a final character at the end of the created file. The
linux utility "cmp" reports this as an "EOF". This creates a problem
for us due to the way we created md5 checksums -- we pipe the output of
"cvs up -p" to md5. Later, when the md5 sum is checked against a file
that was checked out the "normal" way (without -p) the checksum is
different, creating an issue.
How-To-Repeat:
Remove any -k options for a binary file that doesn't have any cvs
keywords. Any cvs keywords will create a corrupted file and will spoil
the test. Do a checkout or update of the file. The do a cvs up -p and
redirect the output to a file like "cvs up -p file > foo". Compare the
two files.
Fix:
Set -kb for the binary file.
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- Difference in output of "cvs up" and "cvs up -p",
Raye Raskin <=