David Glasser wrote:
> On Fri, May 16, 2008 at 11:22 AM, C. Michael Pilato <cmpilato_at_collab.net> wrote:
>> Eric Gillespie wrote:
>>> "Bert Huijben \(@home\)" <bert_at_vmoo.com> writes:
>>>> I don't understand why this value is passed as a value in
>>>> instead of using a argument to svn_client_*() in the c api (We revved all
>>>> these functions in 1.5 anyway).
>>>> It doesn't seem logical that I want to set the same extra revprops on
>>>> command executed during the lifetime of my svn_client_ctx_t. (Many c-api
>>>> users don't recreate the client ctx for every command they use).
>>> I agree this was a mistake; I never use svn_client, so never
>>> noticed. You've caught it too late, I'm afraid.
>> Has he? Seems like a pretty straightforward change to these
>> not-yet-released APIs, don't you think? I'd certainly consider it minor
>> enough to not force a reset of the soak time.
> It's straightforward enough, but will probably cause a reasonable
> amount of (internal, non-rev-requiring) churn in libsvn_client, since
> a whole bunch of functions that already took a svn_client_ctx_t will
> now need a new argument. If people think it would still get
> backported, I'll bang it out.
I would invest time into the review and backport. I'd rather not carry this
wart on ctx_t in this and future releases, basically forcing long-lived
clients to repeatedly set and clear this field around svn_client_* invocations.
C. Michael Pilato <cmpilato_at_collab.net>
CollabNet <> www.collab.net <> Distributed Development On Demand
Received on 2008-05-16 20:32:00 CEST