[Mitgcm-support] RE: Release1 branch

[mitgcm-support at dev.mitgcm.org]

alistair,

 i really don't understand what we are gaining from the new release1 branch?

 it doesn't have the patches we definitely want to carry over that were made
between
 c43 and c44 (about 29,000 lines of diff output!). these were all already
merged into release1-branch.
 they include shapiro updates, changes to GM, fixes for compatability with
TAF etc...

 the things that are in c44 that are not on release1-branch are mostly
tangent-linear
 options. these have not been tested much yet, so they have been left out
conciously. however,
 they could be included if we need them and we can test them.

 branch release1 also doesn't include the tutorial examples.

 to fix these things you're going to have to add the tutorials and merge in
the extra changes.
 both these steps have already been done on relese1-branch and it has
undergone an extensive
 rigourous two-day schedule by hill inc. software testing labs :-). it seems
like a big
 waste of time for you to have to do this again.

 can't i just create a release1-beta2 tag on release1-branch and admit that
 the branch release1 was a dumb and stupid idea. if we want to have a tag
called
 release1-branch-branchpoint on the "MAIN" tree we can do that too.

 it seems like you'll need to spend a lot time just to make it look like
 release1-beta1 was on release1-branch. i don't think anyone will ever care
 either way will they?

chris


-----Original Message-----
From: adcroft at ayana.lcs.mit.edu [mailto:adcroft at ayana.lcs.mit.edu]On
Behalf Of Alistair Adcroft
Sent: Monday, January 28, 2002 12:16 PM
To: support at mitgcm.org
Subject: Re: Release1 branch


Chris Hill wrote:
> I don't think we want tutorial_examples to be part of "checkpoint"
> downloads - do we?

The problem is that the tuts get out of sync with the corresponding
verification. If both tuts and verifs in the development trunk then
we have less of an excuse to change the code in a way that breaks
a tut.

> Also isn't it slightly odd to have a branch called "release1".

Yes but someone had already used up the name "release1-branch" and
I didn't want to break their checked-out codes.

>                                                    That means if
> I type "cvs co -r release1" I get the HEAD of the branch. Is that what
> we want? The original plan was  to create tags on release1-branch called
> ...

Infact, it does work if the branch is never left in an un-tagged state.
But that's not why I did it. I did it because release1 was the only
simple tag available.

> Anyhow - nice diagram, but why didn't you just create a tag
> called relese1-beta2 on release1-branch!

Thanks. Release1-branch already exists and release1-beta2 is
going to contain more than c44.

A.