[ On Thursday, May 11, 2000 at 09:35:49 (-0500), Michael Gersten wrote: ]
> Subject: Re: bug in import?
>
> If I understand this correctly, you are saying to use the trunk for the
> vendor, and branches for local; the opposite of the default (branch for
> vendor, trunk for local).
Well, I'm not saying that you should to things that way -- just that you
can, and that if you do it's more work but perhaps more reliable.
> How do you do the import then? (what is the command for 'take this
> directory tree, stick it on the trunk as the current head, including
> removed or added files or sub-directories'?)
Unless you can find and use my old "import to the trunk" patch there is
no such command. You have to do it by hand, and it's a bit tricky
because the second time around you have to create a working directory
structure that's devoid of files and then unpack the new release into it
and go about adding and removing files and then committing all the
modified files, and of course finally tagging the new vendor release.
In fact even with my "import to the trunk" patch you should probably
still do the second and all subsequent "imports" by hand as I never did
come up with a sure procedure for doing a merge that would remove
deleted files.
> For that matter, what if I wanted to put it on a branch other than the
> vendor branch? I can specify a three digit branch to import, but unless
> I've just done a 'commit -r2', that's not really useful. -- HEY, THAT'S
> A GOOD USE OF REVISION IN CVS! (Comments please, before I shoot myself
> using it)
Only the vendor branch has a three-digit branch number, and you can only
really have one of them (1.1.1).
"normal" branches use magic numbers that you're not supposed to really
try to predict or use externally -- just use the tags.
Without lots of hacking to the internals of "import" you can't use a
"normal" branch either.
> Just as an aside, something that actually came up yesturday; I'm hoping
> that you might have an idea on. We had a source directory that was being
> re-organized on a branch. One of the files was modified on the trunk,
> and had some stuff (two routines) deleted. All of the old stuff was
> empty files on the branch (the re-organization was a complete
> re-structuring of the project).
>
> I was trying to figure out some way to say "Here is the stuff that has
> been added to the trunk since the branch was created, and here is the
> stuff that was deleted from the trunk since the branch was created", in
> the conflict format (easy enough in the patch format, but less useful
> for the cleanup/reorganization
>
> The 'additions' were easy enough, but I still can't figure out how to
> get the deletions into conflict format. Any ideas?
not without using "find", "sort", "uniq", "comm", and perhaps "awk" and
"join"! :-)
--
Greg A. Woods
+1 416 218-0098 VE3TCP <[EMAIL PROTECTED]> <robohack!woods>
Planix, Inc. <[EMAIL PROTECTED]>; Secrets of the Weird <[EMAIL PROTECTED]>
