Mark Phippard wrote:
> On Jan 21, 2008 10:20 AM, Mark Phippard <markphip_at_gmail.com> wrote:
>> On Jan 11, 2008 2:45 PM, David Glasser <glasser_at_davidglasser.net> wrote:
>>> My personal preference would be to merge reintegrate to trunk in the
>>> very near future; then update issue-2897 based on that, then to (on
>>> the 2897 branch) reimplement the FS API without SQLite (I'll do that),
>>> while concurrently folks review Kamesh's libsvn_client changes (not
>>> the backend, which as I said I'll rewrite). Then merge that back to
>>> trunk, then remove the SQLite dependency (which also involves
>>> rewriting the node origin cache on FSFS only, which will be very
>>> easy... just store them as small files in a directory or something),
>>> then BRANCH 1.5.
>> This sounded like a good plan. How are we doing? Step 1 was to get
>> reintegrate merged to trunk. What is left to do for that to happen?
>>
>> Also, are there any opportunities to divide this work up so that it is
>> not all on your shoulders?
>
> Looking at your list again, it looks like a good item that could be
> taken off your shoulders would be rewriting the node origins cache to
> not use SQLite. Could you file an issue or post something to dev@
> that describes how you would do it? Then maybe someone else could do
> that while you are updating the issue-2897 branch?
Node-origins in FSFS can be done by writing a single file for each node-id
which contains the origin value into a 'node-origins' subdirectory. We'd
probably want to shard that in some fashion -- we have around 5000 origins
to store for the Subversion source code repository today.
--
C. Michael Pilato <cmpilato_at_collab.net>
CollabNet <> www.collab.net <> Distributed Development On Demand
Received on 2008-01-21 17:25:44 CET