On Sat, 11 Jan 2003, David Dawes wrote:

>>There are allot of interesting comments here by Mike, particularly his
>>interest in forking XFree86 and creating his own work.  At least I think
>>interesting....and BTW doesn't the ATI maintainer work for Redhat?
>
>Definitely interesting.
>
>One main point I would like to answer now is the issue of the developer
>page on the web site.  New membership was basically closed down temporarily
>while the whole membership/developer situation was being reviewed.
>
>After this message goes out, [EMAIL PROTECTED] will be converted to
>a public mailman list -- so if you want to subscribe, go to
>http://www.xfree86.org/mailman/listinfo/devel/
>
>In the interests of openness, there won't be any private list to replace
>it, and the whole issue of joining XFree86 as a developer will be moot.

David, this is wonderful news.  I'm very pleased to see this 
change happen.  I've wondered for a long time why the private 
devel list was private, considering almost if not all of the 
discussions that have taken place on it in all the time I've been 
on it have not really been something that private.

I hope that the XFree86 core team members will continue to use 
the list now for the same purposes it has been used in the past.  
By active participation between existing core team members, other 
existing member developers, and also new potential developers, it 
may be possible to create a lot more people interested in 
contributing to the project.

So this is very good news.


>BTW, as far as bug tracking goes, there's nothing stopping
>anyone setting up and maintaining a bug tracking system on
>behalf of XFree86.  You really have to understand that forcing
>volunteers to do anything is impossible though, so the
>effectiveness of such a system is something that will only be
>discovered by doing it.

That's always been understood.  I would not want someone to try
and force me to use something any more than one of you developers
on the core team would want someone else to try and force you.

We all of course volunteer our time to work on things that we 
personally feel like working on, and part of that is having 
control of how our own time is spent.  It would be wrong for 
anyone to try to force a developer to use bugzilla against 
his/her will, and would potentially be insulting.


>As far as commit access goes, frankly, if those asking for it
>could establish a record of submitting complete and correct
>patches that didn't need review (and Mike, your record on this
>isn't anything to boast about), then you might have a better
>shot at it.

Well David, out of all of the patches I've submitted to 
XFree86.org that I've written personally myself, the percentage 
of them that were applied to CVS is pretty high, and almost all 
of those without any modification or with minor formatting 
modifications, or a some other similar very trivial change.  When 
patches are committed, before I drop them, I manually examine 
each diff line by line in order to see if the patch was applied 
without modification or not, and to ensure it was done correctly 
without things getting lost.  I also do this detailed patch drop 
review in order to try and understand any modifications that have 
been made to a patch - to learn of course.

In the past I have also submitted various patches found on 
mailing lists, and things people have sent me that fixed 
something for them.  A great deal of said patches I merely passed 
along upstream in hopes that they might be useful to the project 
after someone did a good code review.  This was somewhat of a 
mistake of course, as I had no idea what was expected, and found 
out quite some time later that my processes and procedures for 
submitting patches was not entirely the best.

As you'll recall, as soon as this was brought to my attention, I 
emailed you directly and asked you what you expected in patches, 
and what you expected to accompany them, etc.  I immediately 
modified my methods, and since then, most of the patches I've 
submitted have been of my own writing, or have been much more 
closely scrutinized.  Also, if I can't vouch for a patch's 
correctness, I tend to get the person who wrote it to submit it 
instead, be they some random person off the net, some other X 
developer, or some other Red Hat developer.  I've refused TONNES 
of patches sent to me, and have referred numerous people to 
submit things directly to [EMAIL PROTECTED]  I've also refused 
to apply things until they hit CVS so that I'm guaranteed that 
all people who use XFree86 benefit from a patch, and not just Red 
Hat users.

Other patches I've submitted include various i18n stuff, some 
written by other people at Red Hat, and some of it which was code 
developed by someone off the net, bounced from distro to distro 
to distro and ended up at us, having no idea who originally wrote 
it, and having no familiarity with the area of code it touched, 
but being told that "it fixes Chinese" or similar with no way for 
me to be able to confirm that - as I'm not Chinese.  In these 
cases, I've passed the code upstream to XFree86.org, as I assume 
someone there _is_ able to look at such patches and decide if 
they're crap or not.  And I haven't been let down for the most 
part, because if something is crap, it usually gets rejected, and 
while not always - it is usually rejected with an explanation, 
and such rejections are appreciated.

However, due to fault of my own, I'll be the first to admit that 
I have not had a great track record for patches of unknown origin 
in attempting to find out who wrote them, or even to indicate 
wether I consider the patch good or not.  If I could go back and 
resubmit a lot of patches, I would probably add to a lot of them 
things like:
 
"I didn't write this, and don't know who did.  I was told it 
fixes xyz, but haven't personally confirmed that.  I applied it 
for a test build and it's been in beta testing for x days/weeks, 
and user feedback claims it fixes the problem.  Since I am not 
able to verify the code quality, please review this for possible 
inclusion, and let me know if it is garbage."

If I had been told I was doing something wrong much earlier, I 
probably would have began to change the way I was doing things.  
Until one is told they're doing something wrong however, or 
starts to think it themselves, if they aren't aware of the issue, 
they will probably continue to do it.

This doesn't mean that I've never written a bad patch and sent it 
in, or gotten something wrong the first time however.  Have you? 
Has anyone?  I doubt it.  There aren't that many people who _see_ 
my patches when they're submitted.

I'd certainly be interested in more than one single person doing 
peer review on patches I submit, and then making comments 
publically about my capabilities without any way for anyone 
public to confirm or deny such claims.

I think in the future, I might just post patches here on the 
devel list first to get wider peer review before submitting to 
the patch queue.  That is how the Linux kernel development 
mailing list works, and it seems to work very good.

I very much want to know if something I've written is bad, or has 
a mistake in it, and I'd like to see peer review on a public 
list, one where if you screw up 200 people point it out, not just 
one person.

So please, point out my faults.  My ears are open, and my desire
is great to correct them, and do so in front of my peers.

I'm sure other developers who submit patches would also like to 
have public feedback on what is wrong with them, and to let other 
developers comment on them too.

I eagerly look forward to such open development process.


-- 
Mike A. Harris


_______________________________________________
Devel mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/devel

Reply via email to