Re: rdiff

[Paul Edwards]

"Derek Robert Price" <derek@ximbiot.com> wrote in message 
news:mailman.7166.1054581830.21513.bug-cvs@gnu.org...
> Derek Robert Price wrote:
>
> > Paul Edwards wrote:
> >
> >> So I hope it passes the sanity.sh.
> >
> >
> > I came up with pretty much exactly what you did, aside from some
> > cleanup and additional error messages, and it seems to fix the problem
> > but it seemed to introduce a new one - the tests are failing in
> > head-trunk-diff.  I'm going to try and look into why today.
>
> The problem turned out to be that the code at the end of the function
> was expecting only one or no revision specifying argument.  Once these
> changes were made to remove the DIFF_DIFFERENT return, the functions
> assumptions were blown.
>
> Anyhow, I altered RCS_cmp_file to accept two revisions and to cache the
> first if it checks it out into a file for RCS_exec_diff later.  I'm
> running it through a second sanity check, but I should be checking it in
> shortly.

I noticed the changes on the cvs1-11-x-branch - thanks.

Can you put the phantom "rdiff -s" change on there too?

> No, it was talking about your `id' program being incompatible with the
> script.  It was setting up to match your username in CVS output during
> tests.

I investigated this and found that the man page has a "-u" and "-n"
but not the executable.

The sysadmin reckoned that they must have deleted that option
from the program, and the man page was out of date.  Yeah, right.

BFN.  Paul.