Bug or known weakness in CVS RCS-keyword expansion?

[Peter Toft]

Hi all

If you have two files

a.txt (two lines) : 
--------------------------------
 $Id: $
 $Name: $
--------------------------------

b.txt (one line)
--------------------------------
 $Name: $
--------------------------------

Add those to CVS - check out the module
and run "cvs tag TAG1"

Then 
 $ echo "hi there" >> a.txt
 $ cvs commit -m "hi added" a.txt
 $ cvs tag TAG2

If I then run
 $ rm a.txt b.txt
 $ cvs update -r TAG2

Then I see that the label "TAG2" is inserted after the magic keyword 
"Name" in both files.

If I now continue running
 $ cvs update -r TAG1
Then I see that "a.txt" is updated with version and TAG1 info - however 
"b.txt" still reads this line only
 $Name: TAG2 $
since TAG1 and TAG2 for the b.txt is the same version.

However if I do
 $ rm a.txt b.txt
 $ cvs update -r TAG2

Then both files will have the wanted TAG2 inserted.
I think the current behaviour is unfortunate - but is this a bug? 
Should I report it as a bug?

I am also thinking about this in the context of the CVS bug I found 
recently
http://savannah.nongnu.org/bugs/index.php?func=detailitem&item_id=17032
where I think (read comment #3 from Jim Hyslop) that the MD5 must in both 
sides use the unexpanded keywords - otherwise things go bad (especially 
given what I explain above).

Comments please?


Peter Toft, Ph.D. address@hidden http://pto.linux.dk

"First they ignore you, then they ridicule you, then they fight you, then you 
win."
  --  Mahatma Gandhi