Package review backlog.

[Thorsten Leemhuis]

On 10.10.2008 15:14, David Woodhouse wrote:
> On Fri, 2008-10-10 at 02:12 +0200, Christoph Wickert wrote:
>> Am Donnerstag, den 09.10.2008, 20:01 -0400 schrieb Josh Boyer:
>>> On Fri, Oct 10, 2008 at 01:41:32AM +0200, Christoph Wickert wrote:
>>>>> Let's get the final two packages reviewed -- and that's another area
>>>>> where we could do with some improvement, because failing to approve
>>>>> packages really _is_ verging on the 'deletionism' you spoke of. But
>>>>> that's a separate discussion.
>>>> Agreed, we can discuss this later. BTW: The word "deletionsm" did not
>>>> come from me, it was Josh who said that, but of course I agree with him.
>>> I didn't say that.  Though I agree we need to be getting more packages
>>> approved.
>> I think this is something we all agree on. ;)
> I propose that each FESCo member should try to work on at least one
> package review per week. Each week at the FESCo meeting, we'll ask
> members which reviews they've worked on in the past week.

I blogged about this already(¹), but a few hours later I think I should 
write a alternate proposal to prevent the problems I tried to outline in
https://www.redhat.com/archives/fedora-devel-list/2008-October/msg00560.html

So here we go:

---

I propose that each week at the FESCo meeting all FESCo members should 
be asked what they did (as FESCo member) over the past week to improve 
Fedora. They further should be asked what they (as FESCo member) plan 
for the coming week to improve Fedora.

---

Do I mean that serious? If the proposal from dwmw2 is considered to be 
accepted, then yes, definitely. In fact it might be wise to combine 
those two to something like that below *if* FESCo really wants to go 
down that route (which I still think is totally wrong to do, but that is 
a different topic; I'm just trying to make something that IMHO is worse 
a bit bitter):

---

I propose that each FESCo member should try to work on at least one 
package review per week. Each week at the FESCo meeting all FESCo 
members should be asked what they did (as FESCo member) over the past 
week to improve Fedora; they further should be asked what they (as FESCo 
member) plan for the coming week to improve Fedora. That includes, but 
is not limited to reviews.

---

That way I as ordinary contributor know way better how the FESCo member 
users his time. I then are able to say "okay, he did not do any review 
over the past month week, but I can see that he spend his time for 
things that were more important and overall improve Fedora; that is way 
better than one or two finished reviews". I can use that information in 
the next election.

CU
knurd

(¹) see
http://thorstenl.blogspot.com/2008/10/popular-things-are-often-not-right.html
or just read on:

Popular things are often not the right thing to do

No, this has nothing to do with the presidential elections in the US; 
but yes, it's about politics. To quote from <a 
href="https://www.redhat.com/archives/fedora-devel-list/2008-October/msg00673.html">a 
mail send by dwmw2 to fedora-devel</a>:

> I propose that each FESCo member should try to work on at least one
> package review per week. Each week at the FESCo meeting, we'll ask
> members which reviews they've worked on in the past week.

Nice idea, but for me that sounds like a pre-election promise to lower 
taxes or like big and rich politicians handling out food in a soup 
kitchen or a: It will help a few people and gives good press because 
it's popular, but it doesn't solve the underlying problem at all. In 
fact it's even worse, because time gets wasted instead of solving the 
underlying or other problems (that might be bigger or smaller).

Verbose:

Sure, putting pressure on FESCo members to do reviews will force them to 
face some of the problems in the review process. That might result in 
some small improvements to the review process that might make things 
easier in the long term (but for that to really become true we also 
would need to have the same "should try [...] at least one package 
review per week" suggestion for members of the Fedora Packaging 
Committee as well); and of course the review queue gets a tiny fraction 
shorter due to the reviews that get done. But the time imho would be 
*way* better spend if some FESCo members would directly work on 
improving the review process with people that are doing lots of reviews 
(hello tibbs), because only that will improve the situation in the end 
(if done properly) and solve the problem (as far as it's possible to solve).

Ideas for making things easier and better are there; they are afaics 
well known among contributors and the different committees that are 
responsible. Just nobody is working on them:

  * improve rpmlint and other tools to automate more of the checks to 
less the burden on the reviewer

  * make review exchanges easier; maybe even force/guide/direct people a 
little bit to do exchange reviews (e.g. a little bit like, but not as 
strict as "if you want to get your packages reviewed you have to review 
a package from somebody else first")

  * look more actively for new sponsors and be less strict when choosing 
them (like it has been in the past); we all make errors -- it's the 
ability to learn from them and to fix the errors once they got made

  * be less strict with sponsoring people like it was in the early 
Fedora Extras days in 2005. We can do that now that new packagers don't 
get access each and every package in CVS

  * add one more level between new packagers and sponsors; soon-to be 
sponsors there could work together with new packagers and keep an closer 
eye on them; by that they could get work of the real sponsor and show 
their ability to become a real sponsor sooner or later

  * let FESCo and the Fedora Packaging Committee work together to make 
review and packaging guidelines easier to understand. We have done that 
two times in the past iirc (once in the fedora.us days and the last time 
it iirc mainly was spot's work with some help from mschwendt, scop and 
some others in the early FESCo days before the FPC existed). I tend to 
say it's overdue to do it again. Guidelines for corner cases in that 
process should get moved to special add-on documents or sections that 
are hidden by default. That will make the main things easier to 
understand and remember. Otherwise we soon have guidelines that will 
look like a code of law/statute book that nobody really understands as 
knows, as they are long and quite hard to read. Maybe splitting the 
guidelines might make sense as well: a "this is how it works in general" 
  could be the quick ans easy start; a "here is how it works in detail" 
could serve as reference doc wher you have to look for the details and 
special treatments when it comes to perl/python/mono/java/...

  * there are likely more ideas floating around...

And note, package reviews are just a fraction of the the area that FESCo 
is responsible for. So in the end instead of wasting time in each FESCo 
meeting with questions like "which reviews did you work on over the past 
week" it would imho be way better to ask "what did you do over the past 
week to make Fedora (the distribution and the project) a better place 
for contributors and users". That would also help to answer the "whom to 
vote for in the next election" question a lot.