Bug#749497: validation
Hi Adam & William,
On Thu, Jun 19, 2014, at 9:29, William Dauchy wrote:
> Hi Adam,
>
> On Wed, Jun 18, 2014 at 3:01 PM, Adam D. Barratt
> <adam@adam-barratt.org.uk> wrote:
> > "Simple backports from upstream" is not equivalent to "suitable for an
> > update to a package in stable", particularly when there's a non-trivial diff
> > involved.
>
> All the backported fixes are bugs which make the package more stable.
> It fixes real issues we have in php5.4.
>
> > And then another and another and... ? My concern here is that we end up with
> > a continual stream of requests for updates, each with many fixes in. I'm not
> > sure how sustainable that is; ommv, of course.
The PHP code is fragile and plagued by many bugs that ends up
in crashes. If we are to improve the overall stability of PHP packages
in Debian (so the upstream won't tell the users - compile the code
yourself), we inevitably end up in stream of requests for updates.
Our goal is not to push just some random upstream updates, but
to provide better experience for Debian PHP users.
We just need to find a model that would work for both parties. I
understand
that reviewing PHP patches is a big burden, but on the other hand it's
the
PHP team responsibility to provide a stable p-u that has received
extensive
review.
So it's not the question how many updates there will be, but the
question
if you believe the team that you would be able to trust us that we don't
fuck up each update and that we won't abuse it in the future.
I think we need to cut this quickly so neither side gets frustrated and
And I see that we are heading that way even though there's a good
will at either side. And that worries me much.
Ondrej
--
Ondřej Surý <ondrej@sury.org>
Knot DNS (https://www.knot-dns.cz/) – a high-performance DNS server
Reply to: