Re: [RFA] convert blocks to dictionaries, phase 1, main part

[David Carlton <carlton@math.stanford.edu>]

On Tue, 17 Sep 2002 13:49:28 -0400, Daniel Jacobowitz <drow@mvista.com> said:
> On Tue, Sep 17, 2002 at 10:43:42AM -0700, David Carlton wrote:

>> Is it okay to put up an RFA after I've switched over just blocks,
>> or do you want me to wait until after I've switched over global
>> symbols as well?

> Basically, at any point when you don't have a lot of temporary gunk.

Great; I'll get rid of all the DICT_TEMP gunk, finish converting
blocks over, and send out a new RFA.  That won't take a separate
branch.  And maybe before doing that I'll send out an RFA that gets
rid of BLOCK_SHOULD_SORT.

> How about a branch which require approval just like the mainline for
> large patches, although giving David a little more freedom to play
> around.  Then, we'd allow large merges from the branch back to the
> trunk when they were ready and tested - larger patches than we'd
> normally accept all at once, because they'd already been approved.

From my point of view, having a separate branch could be quite useful
when getting to the global symbols, and what you propose could be a
reasonable mechanism.  For one thing, it seems plausible to me that it
might take some amount of playing around with implementation methods
before settling on one that works well, so a branch is a natural place
for that sort of playing around to occur.  And, for another thing,
having it be in a branch increases the chances that people other than
me can be shanghaied into working on this.  (I can work on converting
appropriate code to using struct dictionaries, you can play around
with getting the interface to symbol lookups just right, and if we're
lucky Daniel Berlin can pop by every once in a while saying that he
took a break over the weekend from law school stuff to tear out all of
the psymtab->symtab translation and replace it with a new, improved
mechanism, or something like that.)

So I'm not resistant to branches generally.

Incidentally, I really should send out some e-mail that talks about
what I think the eventual interface to symbol lookups should be.

David Carlton
carlton@math.stanford.edu