19 Feb
2019
19 Feb
'19
9:48 a.m.
Hi Simon,
On 19 Feb 2019, at 08:16, Simon Urli
<simon.urli(a)xwiki.com> wrote:
Hi Vincent,
On 16/02/2019 12:25, Vincent Massol wrote:
Sure, we can have a look at the bugs existing in jira and try to take the “more important”
one first. However be careful that the severity in JIRA may not reflect what we think. Any
user having an issue that is blocking him/her will consider it critical even though it may
be a special use case that we consider a not important at all compared to other more
important bugs.
Also note that 1-2 days to fix a bug is quite large and show allow to fix a big majority
of them.
Last point is that we’re not waiting for bug fix releases to fix important bugs, we do
that every day already and that’s good. The point here is not just to do more of what we
usually do, it’s also to do what we don’t do often (we do that once per week during BFDs
only), i.e. fix easy bugs and also take the time to clean up our JIRA by closing bugs we
won’t fix, that are duplicates, etc. Said differently closing 5 bugs that are marked
‘major’ can be a lot better for XWiki than fixing one hard bug marked critical but that’s
specific to a seldom-used use case.
I’d still like that we keep the spirit of trying to close the maximum number of bugs and
stay within the 1-2 days if we can, as the general rule. Remember that the reason we’re
doing these 2 BFD releases is to close more bugs than there are bugs created over the past
1600 days and go back to 0, which is the situation we were in, about 1-2 years ago. Thus
we’ve lost ground and the goal is to catch up.
Of course, if we notice an important bug, we can quickly discuss it between ourselves and
decide if it’s really important and fix it. If you already have some ideas of such bugs,
let’s discuss them.
Thanks
-Vincent
Hi devs,
No feedback so I guess everyone is ok with this.
I’ve now updated https://www.xwiki.org/xwiki/bin/view/Roadmaps/
I’ve also suggested to have an objective of collectively reducing the bug count by 50 in
XWiki 11.2 and 50 more in XWiki 11.3, with the goals of closing bugs (not necessarily
fixing them, BFD-like).
My computation: 1 bug closed per day (we can easily close 5-6 if we find dups/won’t
fix/etc, taking 1 to have margin) * 5 (days of week) * 3.5 (for the month, keeping some
margin) * 3 (active FTE devs on XS) = 52.5.
Ideally we should focus on quantity and not quality, and take bugs that don’t take more
than 1-2 days to fix.
I'm mixed with this idea of focusing on quantity and not quality here: when it's
about BFD I'm ok with it since we only got one day to fix them. Now here we got two
months, so we can use them to also focus on more difficult bugs that are there for a
while, at least to discuss on what we plan to do to fix them.
To give some figures, out of 44 critical bugs in Commons/Rendering/Platform 31 of them
are there for more than one year. I know those are pretty difficult to fix, but then we
can spent a bit of time on them to at least analyze them.
Now maybe I'm just too young on the project and all of them have been thoroughly
analyzed already and they're not fixed because basically we cannot right now?
WDYT? Simon
That seems doable to me, WDYT?
Thanks
-Vincent
> On 11 Feb 2019, at 16:33, Vincent Massol <vincent(a)massol.net> wrote:
>
> Hi devs,
>
> We have already started defining the roadmap for XS 10.2, see
https://www.xwiki.org/xwiki/bin/view/Roadmaps/#HXWiki11.2
>
> However, it’s come to my realization that we’re having a lot of stability issue on XS
these days. I see plenty of users messages on the forum about users failing to use XWiki
or to upgrade XWiki and having problems in general. I have the feeling that our quality
has regressed.
>
> If we check our number of bugs created vs number of bugs closed for 2018, we can
clearly see that we’re loosing the battle since Feb/March 2018, see:
>
https://jira.xwiki.org/secure/Dashboard.jspa?selectPageId=10352#Created-vs-…
>
> Since Feb/March we’re at 820 vs 963 which means we have a relative difference of 143
bugs. Note that the BFD sessions are not enough and they don’t close a lot of bugs anymore
(this can be seen on https://www.xwiki.org/xwiki/bin/view/Blog/XWiki%20Days).
>
> I believe that XWiki’s stability and reducing user frustration for
installation/upgrades is a key aspect and it’s one of the biggest possible contributor to
Active Installs.
>
> Thus I’m proposing to devote 2 releases (i.e. 2 months) to focus on bug fixing and
thus on XWiki stability, namely
> * XS 10.2
> * XS 10.3
>
> Note that this is something that I’ve already discussed with committers from XWiki
SAS. Please voice your opinion if you disagree (or even if you agree, always good to get
confirmation we’re going in the right direction ;)).
>
> Thanks
> -Vincent
>
--
Simon Urli
Software Engineer at XWiki SAS
simon.urli(a)xwiki.com
More about us at http://www.xwiki.com