Linux Archive

Linux Archive (http://www.linux-archive.org/)
-   Gentoo Development (http://www.linux-archive.org/gentoo-development/)
-   -   Regarding the tinderbox logs (http://www.linux-archive.org/gentoo-development/709343-regarding-tinderbox-logs.html)

Diego Elio Pettenò 10-03-2012 02:35 PM

Regarding the tinderbox logs
 
On 02/10/2012 21:14, Ryan Hill wrote:
> Well, duh. You designed, developed, and are the sole architect of the
> system.

Not by choice...

> You made an error in the design. You might have had good reasons at
> the time, and you can argue them til you turn blue to anyone who will listen,
> but if your end consumers see it as a flaw it isn't going to change
> a thing.

Here's the problem, it's the 80-20 rule. Just in this case reversed, in
the sense that 80% of the noise comes from 20% of the people (and I'd
argue even less than 20%).

We have a system that works nicely, and most people don't even complain
about it. A few asked why, and when they were told they said "ok". A
couple complained that it's not as easy but kept going and _one_ is
destructively removing references to the logs and ignoring valid bugs on
his own packages (caused by his own packages most of the time), because
they smell.

It does seem logical that I'm not going to rollback months of work just
because one guy can't be bothered to play well with others, doesn't it?

--
Diego Elio Pettenò — Flameeyes
flameeyes@flameeyes.eu — http://blog.flameeyes.eu/

Markos Chandras 10-10-2012 08:47 AM

Regarding the tinderbox logs
 
On Wed, Oct 3, 2012 at 3:35 PM, Diego Elio Pettenò
<flameeyes@flameeyes.eu> wrote:
> On 02/10/2012 21:14, Ryan Hill wrote:
>> Well, duh. You designed, developed, and are the sole architect of the
>> system.
>
> Not by choice...
>
>> You made an error in the design. You might have had good reasons at
>> the time, and you can argue them til you turn blue to anyone who will listen,
>> but if your end consumers see it as a flaw it isn't going to change
>> a thing.
>
> Here's the problem, it's the 80-20 rule. Just in this case reversed, in
> the sense that 80% of the noise comes from 20% of the people (and I'd
> argue even less than 20%).
>
> We have a system that works nicely, and most people don't even complain
> about it. A few asked why, and when they were told they said "ok". A
> couple complained that it's not as easy but kept going and _one_ is
> destructively removing references to the logs and ignoring valid bugs on
> his own packages (caused by his own packages most of the time), because
> they smell.
>
> It does seem logical that I'm not going to rollback months of work just
> because one guy can't be bothered to play well with others, doesn't it?
>
> --
> Diego Elio Pettenò — Flameeyes
> flameeyes@flameeyes.eu — http://blog.flameeyes.eu/
>

Nobody asked you to rollback months of work just because a few people
can't deal with the way you submit your logs. But you should also
respect their preferences. Each one of us has its personal way of
dealing with his/her bugs (based on some loose standards and
policies). If someone doesn't like your way of submitting logs then
just accept it. It is not like they go and remove your tinderbox links
from bugs/packages they don't maintain.

--
Regards,
Markos Chandras / Gentoo Linux Developer / Key ID: B4AFF2C2

Diego Elio Pettenò 10-10-2012 02:41 PM

Regarding the tinderbox logs
 
> Nobody asked you to rollback months of work just because a few people
> can't deal with the way you submit your logs.

Actually somebody did, suggesting I shouldn't file bugs until the log's
attached. And then proceeded to suggest that converting everything to
python and using pybugz is a cakewalk (obviosuly without offering to
even start the work).

> But you should also
> respect their preferences. Each one of us has its personal way of
> dealing with his/her bugs (based on some loose standards and
> policies).

Sure. Preferences are great. Until said preferences mean that bugs that
_are_ 100% valid get closed, repeatedly, without being looked at.

Let's make an example. In the recent past, an update to binutils broke
some of the libbfd (the library coming with it) users as headers were
changed around. How many copies of the same bug do you think one has to
file before policies should overrule preferences?
(and this is not a matter of just refusing the log, it's refusing the
_bug_ which is extremely easy to reproduce).

> If someone doesn't like your way of submitting logs then
> just accept it. It is not like they go and remove your tinderbox links
> from bugs/packages they don't maintain.

Actually, that happened as well. Maybe you should actually review facts
before posting sure that you know that's going on. Just saying.

--
Diego Elio Pettenò — Flameeyes
flameeyes@flameeyes.eu — http://blog.flameeyes.eu/

Rich Freeman 10-10-2012 03:18 PM

Regarding the tinderbox logs
 
On Wed, Oct 10, 2012 at 10:41 AM, Diego Elio Petten
<flameeyes@flameeyes.eu> wrote:
> Sure. Preferences are great. Until said preferences mean that bugs that
> _are_ 100% valid get closed, repeatedly, without being looked at.

I can't speak to the specifics of whatever the elephant in the room
is, but keep in mind that when you have 100 developers on a project
there is only so much you can tailor things to individual preferences.

I've gotten tinderbox bugs that are upstream issues that I personally
accept as valid but which I'll probably never be able to influence
upstream to change. I just read them, appreciate them, and then leave
them open in the hope that I'll be bored one weekend and find some way
to fix it. Usually I find Diego's bugs to be helpfully worded both
with helpful logs and background info which saves me having to explore
some arcane linking issue.

If the result of a tinderbox run is we get 500 bugs that are legit, 3
false positives that waste somebody's time, and 5 legit but
debate-ably unimportant bugs that particular maintainers don't want to
look at, I'd say we came out ahead.

This could be a culture thing. At work I tend to work with large
regulated applications that often have hundreds of open bugs at any
time - some for years with no intention whatsoever to actually close
them. Bug lists aren't really used like worklists in this context,
except for the few times a year everybody sits down and prioritizes
the list and figures out what is worth fixing, if anything. So,
having a few open bugs assigned to me doesn't really bother me.

Maybe the solution is some kind of maintainer priority field for
personal productivity which is to be set by maintainers only and can
be filtered on if a maintainer just doesn't like seeing things in
their daily view.

Rich

Markos Chandras 10-10-2012 03:28 PM

Regarding the tinderbox logs
 
On Wed, Oct 10, 2012 at 3:41 PM, Diego Elio Petten
<flameeyes@flameeyes.eu> wrote:
> Actually, that happened as well. Maybe you should actually review facts
> before posting sure that you know that's going on. Just saying.
>
Ok

--
Regards,
Markos Chandras / Gentoo Linux Developer / Key ID: B4AFF2C2


All times are GMT. The time now is 11:04 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.