On 19.08.2019 15:58, Julian Foad wrote:
> Brane, do I correctly understand the impact of this branch could be
> summarized as "API support to enable a future version of the client
> software to support multiple WC versions":
>
> * Enables future client software to detect, create, use without
> upgrading, and optionally upgrade any WC version in the supported range.
>
> * The supported range is initially just the current WC version; this
> does not add a new WC version to the supported range.
>
> * To actually support multiple WC versions, the client software will
> need explicit cases for each supported WC version in the handling of
> each feature that varies across versions.
Yes, this is all correct.
My initial plan was to implement support for compresseed/BLOB pristines
after this support is on trunk. But that went the way of so many plans,
i.e., into the no-time zone. :/
>
> That seems to match the branch description at
>
> http://svn.apache.org/repos/asf/subversion/branches/multi-wc-format/BRANCH-README
>
>
> Julian Foad wrote:
>> [...] > I assume Brane has tested it, but someone else testing too
>> would be
> good and I haven't. Then this looks likely to be ready to merge to trunk.
>
> Thinking again, we need an implementation of multiple WC versions to
> test it. It would be wise not to merge until we have tested with such
> an implementation.
Agreed. That's why I only asked for a review, to add eyes and see if I
was going in the right direction. There's some support on the branch to
actually test the multi-format support, but IIRC it sort of trips on a
hard-coded #define -- I'd have to look into that again, it's been a long
time.
> What is the most feasible way forward with this? Is the
> 'better-pristines' branch anywhere near providing at least "dummy"
> handling of a new WC format, sufficient to test these support APIs,
> even if it is not yet actually implementing better-pristines?
As I said, that was the plan, but it's not too close yet.
-- Brane
Received on 2019-08-19 16:27:13 CEST