RE: Proposed branch tag performance patch for feature

[Kelly F. Hickel]

Hi Mark,

                I've got to run for the day, but I wanted to get back to
you, I'll look at this a bit more tomorrow.

<snip>
>Hmmm... My test had a problems with a number of the branches3 tests...
>The first was branches3-4b.
 I took a quick look at this (unzipped the original code, applied your
patch and ran the tests).  I'm not sure this really explains the issue,
but the sanity test seems to have changed, where my patch had:

+         dotest branches3-4a "${testcvs} -q tag -b br3br1" 'T file1
+T file2'
+         dotest branches3-4b "${testcvs} -q update -r br3br1" ''


Yours had:
+         dotest branches3-4a "${testcvs} -q tag -b br3br1" \
+'T file1
+T file2'
+         dotest branches3-4b "${testcvs} -q update -r br3br1" \
+'[UP] file1
+[UP] file2'

Branches3-4b doesn't product any output in my tests, if I change the
expected text for that one back to '', the test succeeds.

<snip>
> 
> In cvs 1.11.x, we use K&R declarations everywhere.
> 
>   static void
>   findnextmagicrev_delproc(p)
>       Node *p;
> 
> in cvs 1.12.x, we should be using ANSI C89
> prototypes everywhere...
[Kelly F. Hickel] Yeah, that one slipped through somehow....
<snip>
> > +  info.target_rev_dots = numdots (rev) + 2;
> > +  info.target_rev_len = strlen(rev);
> > +  info.rev_list = getlist();
> 
> The coding standard wants spaces between function
> names and argument lists:
> 
>   info.target_rev_len = strlen (rev);
>   info.rev_list = getlist ();
> 
> Generally, there are style problems thoughout this function.
> 
[Kelly F. Hickel] Yes, I saw that early on and tried to keep it in mind,
but I suspect that I forgot it as soon as I started working on it the
next day.
<snip>


Let me know how you want me to proceed, I can take start with our patch
and get the tests working again, if you like.

Thanks,
Kelly