Greg Stein wrote:
> On Wed, Oct 14, 2009 at 03:38, Branko Cibej <brane_at_xbc.nu> wrote:
>
>> Paul Querna wrote on dev_at_apr:
>>
>>> fixed, it was due to a problem with svn:externals values changing.
>>> looking at a long term solution.... but not sure if there is a good
>>> one.
>>>
>>>
>> I should be collecting such testimonials. I opposed svn:externals in
>> their current form from day 1, on the grounds that such an 'orrible hack
>> would cause no end of trouble. Vindication FTW ...
>>
>> (This was a "change an svn:externals URL on one client and svn:up on
>> another, bingo, the change isn't reflected in the working copy contents")
>>
>> /me breathlessly waits for the standard "just another bug to fix" response
>>
>> </rant>
>>
>
> It was a replacement for CVS's "modules" concept. You never suggested
> an alternative, that I recall...
>
I'd have to dig into old archives, but I'm sure I suggested we make
aliases a server-side feature, i.e., that the client should not know how
the server internally remaps the paths; sort of like a server-side
symlink. ISTR the counter-argument was that doing this server-side would
be a big job that would delay 1.0, and that we couldn't easily support
externals referring to other repositories that way (no matter that CVS
can't either, so the feature parity argument is on shaky ground).
With one thing leading to another, I never got around to implementing my
idea, and avoid using externals for my own purposes as far as possible.
I still feel the old rant rising up, though, every time I see some
problem reported with regard to externals.
Anyway, yeah, I do rant and bitch about this on a moderately regular
basis. I should just learn to stop, since no-one appears to agree with
me (except for the people who report the problems in the first place).
-- Brane
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2407684
Received on 2009-10-14 20:31:05 CEST