Re: [RFC] Mergeinfo and missing children of merge target

Paul Burba wrote:
>> Rule #1: a revision range prefixed by a hyphen is negated
>> Rule #2: no uninheritable ranges may be syntactically optimized away
>
> I don't understand what Rule #2 is about, specifically what
> "syntactically optimized away" means. Could you give an example or try
> and alternate explanation?

Oh. I mean simply that you aren't permitted to reduce "/A:1,3-4,-3-4*" to
just "/A:1" as a storage optimization. Is this what you guys are calling
"eliding"?

>> Using negated merge ranges, you could store:
>>
>> "/A:1" on A_COPY and all directories beneath it for whom every
>> child is present.
>
> When you say "and all directories" you don't mean explicitly right, but
> via mergeinfo inheritence I assume?

Right. Sorry.

>> "/A:1,3-4,-3-4*" on any directory FOO under A_COPY whose children
>> are not all present (A_COPY\D\H, in your example).
>>
>> "/A:1" on children of FOO which are present.
>
> a.k.a. The present siblings of the missing path get inheritable
> mergeinfo.

Exactly.

-- 
C. Michael Pilato <cmpilato@collab.net>
CollabNet   <>   www.collab.net   <>   Distributed Development On Demand