Bug 71264

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20020625

Description of problem:
If the hook function po-replace-revision-date is disabled by setting the
variable po-auto-replace-revision-date to nil (which is a legal value according
to the documentation), it is no longer possible to save the po file.

Version-Release number of selected component (if applicable):21.2-10


How reproducible:
Always

Steps to Reproduce:
1.Start emacs on some po file: emacs x.po

2.Set the variable to nil
M-x s e t - v a r i a b l e <return> <help-echo> p 
o - a u t o - r e p l a c e - r e v i s i o n - d a 
t e <return> n i l <return>

3.Do some editing of any message in the file.

4.Try to save (or verify) the file.

Actual Results:  The file isn't saved.  If verifying, the original file is
checked.  If quitting po-mode, a second question comes up asking if the buffer
should get killed although it's modified.

Expected Results:  The file should have been saved.

Additional info:

In the file po-mode.el, the hook function po-replace-revision-date is 
automatically added to write-contents-hook.  Normally it updates the
PO-Revision-Date.  It is configurable, though, through the variable
po-auto-replace-revision-date.  According to the documentation, it can be nil,
t, or ask, and is t by default.

If this variable is nil, I believe the intention is to simply disable  this
function.  But the function in this case ends with a call to the function
message with the argument "".  The return value of this call is also "", which
becomes the value of the complete hook function.

When the hooks in write-contents-hook is called, a non-nil value from any hook
function is interpreted to mean that the file is already written.  The net
effect is that the po file never is written.

I believe this is what would happen if po-auto-replace-revision-date is set to
ask, and the question is asked with no, or if it is called somewhere where
format-time-string is not available.  I've not verified those last two
assumptions though.

I've reported this error upstreams.  I'm submitting this bug in case you want to
patch your distribution until it is fixed officially.