2024-01-03 17:44 AEST by Arthur Barrett - after commit, the local file is updated.  But why?  And why for a binary?


eg: this error report:

In P:\CVS\ES1912: "C:\Program Files (x86)\CVS Suite\CVSNT\cvs.exe" -q commit -m "ES1912-PlayPen: 
Spelling error correction in Propagation.m" PlayPen/timg/L1Analysis/AGC/Propagation.m 
PlayPen/timg/L1Analysis/AGC/Propagation.png
CVSROOT=:ssh:timg@cvs.indianpacific.com.au:/cvs

/cvs/ES1912/PlayPen/timg/L1Analysis/AGC/Propagation.m,v  <--  PlayPen/timg/L1Analysis/AGC/Propagation.m
new revision: 1.2; previous revision: 1.1
/cvs/ES1912/PlayPen/timg/L1Analysis/AGC/Propagation.png,v  <--  
PlayPen/timg/L1Analysis/AGC/Propagation.png
new revision: 1.2; previous revision: 1.1
*******************************************************************************
** THIS ERROR SHOULD NEVER OCCUR - ALWAYS REPORT IT TO:
**                          support@march-hare.com
** 
** Unable to rename "CVS/Base/_new_Propagation.png" 
**        rename to "Propagation.png".
** 
**   CLIENT CVSNT 2.8.01 (Soolin) Build 8078
** 
.
.
.
** 
** Attempt to delete/move file using the Shell:
** Delete file using the Shell was successful and not aborted!
** Renamed file Propagation.png to CVS/Base/CVS/_new_Propagation.png4db46594da8c6915 fails.
** 
** Unable to rename "CVS/Base/_new_Propagation.png" 
**        rename to "Propagation.png".
** 
** THIS ERROR SHOULD NEVER OCCUR - ALWAYS SEND THE ABOVE DIAGNOSTIC TO:
**                          support@march-hare.com
**   CVSNT 2.8.01 (Soolin) Build 8078
** 
*******************************************************************************
cvs commit: 0x5 Access is denied.

cvs [commit aborted]: [14:55:23] cannot rename file CVS/Base/_new_Propagation.png to Propagation.png: 
Permission denied

Error, CVS operation failed (exit code 1)




Typically this error happens due to anti-virus.  But in this case it's probably just the editor still had the file 
open and the customer wouldn't expect "commit" to modify the file.


I think it's a reasonable expectation.

It's not like the file will have keywords that need updating...


Can't we just skip this for binary files and files without keyword flags (-ko).

Skipping for files with keywords -kkv but no actual keywords in the file would probably be nice too.