On Wed, May 27, 2020 at 8:06 PM Jeff King <peff@xxxxxxxx> wrote:
>
> On Wed, May 27, 2020 at 06:47:32PM +0200, Christian Couder wrote:
> > They are only used by protocol v0 code since protocol v2 assumes
> > certain baseline capabilities, but rolling them into
> > upload_pack_data and just letting v2 code ignore them as it does
> > now is more coherent and cleaner.
>
> Is it perhaps worth keeping these v0-only fields grouped together within
> the struct, along with a comment?
Ok, I will try to do that in part 2 and part 3.
> That way nobody has to repeat your
> research later into which ones are used and which ones are not.
>
> > -static unsigned int timeout;
>
> This was unsigned, but gets moved as...
>
> > @@ -83,18 +80,21 @@ struct upload_pack_data {
> > timestamp_t deepen_since;
> > int deepen_rev_list;
> > int deepen_relative;
> > + int timeout;
>
> ...a signed int.
Yeah, not sure what happened. Maybe I needed some coffee when I did that.
It's fixed in my current version.
Thanks,
Christian.
