[cvsnt] cvs update does not set ErrorLevel

[Gerhard Fiedler]

Michel Terrisse wrote:

> So the exit code is not reliable. 

I have found that, too. It can be surprising...

> Why does cvsnt return 0 when it fails ? 

I think one problem may be that "failure" is different depending on what
you are looking for. Rather than a two-value "fail/ok" result, a more
elaborate result code might be required, and failure would then be defined
as greater than a certain threshold. Or maybe even a bitmap rather than
various levels.

> Is there any way to detect that the command has failed ?

I think the only way is to parse the textual output. You'd have to
implement a more fine-grained result code in a script.

Gerhard