Re: Assert in recurse.c when 'checkout -p' used

[Doug carr]

Hi Larry,

>> Just trying to decide if it is
>> worthwhile to change my scripts and upgrade to 1.11.10 (with the danger being
>> that that message will change when the assert is fixed, requiring another
script
>> change), or be patient and wait for a patch so I don't have to change my
scripts
>> (assuming the error message is changed back to 'no such tag X')...

>That would be the correct error message.  Can't you just change your
>scripts to allow either message?  :-)

Yeah, that's what I've done for now.  My worry is another version of that
message will appear in the next release, and then I end up with three things to
check, etc.

A bigger worry is that there are other spots in the code that also issue
different messages than what I was expecting.  Ideally, each error string issues
by CVS would have a [Exxxx] in it so I could just match on an error number
rather then depending on spelling, language,... being consistent.  An added
benefit would be that a list of those could be published so I know I've covered
all the possibilities without trying to find all the strings in the source code.
It should probably be enabled via a configuration option so people who depend on
the undecorated strings (like me with my current scripts) wouldn't complain
about the change in the strings...

Thanks again,
Doug.