Maciej Stachowiak 2009-02-02 12.21:
On Feb 1, 2009, at 10:00 PM, Leif Halvard Silli wrote:
Ok, Maciej, thank you for saying so. [...]
I will add that I agree with your core point that issues should be
clearly recorded. I think the combination of bugzilla and the wiki
tracker, with occasional spec notes, are OK for this, but I think
reasonable people can disagree.
My thinking on this is colored by my experience as a software engineer.
For software products, it is generally thought that defect reports are
best recorded in a separate database which can be searched and queried.
Occasionally, it is helpful to record an open issue in the source code,
but this is done rarely. The overhead of recording all issues this way
would be large, and it would be difficult to search or query the
resulting records. To me, a source code comment corresponds to an HTML
comment note in the spec. Very rarely, a defect is noted in the visible
UI of the software. This is only ever done for seriously unfinished
parts of unfinished software products. To me, this is what corresponds
to a visible note in the spec.
>
I can understand that people might have a different perspective on this,
and I will myself admit that a standards document is not exactly the
same kind of thing as a computer program.
I do think this group hasn't done such a great job of tracking issues,
but I think adding bugzilla to the mix has certainly helped.
I also think it would be useful to have this discussion about ways to
improve issue tracking on the working group list. I'm sure lots of
people have good thoughts on the topic. But I leave that to Sam's judgment.
Thank you for these comments. I'm glad we agree on the need to
track issues better.
This is the issue in Sam's message that we debate:
3) Members of the working group may raise issues at any time.
Issues are to be raised on the public-html mailing list, need
to be specific and contain an actual proposal to be considered
OPEN, and will be indexed by the
http://www.w3.org/html/wg/tracker/issues list. Until resolved,
if three independent members of the working group agree that
this is an issue, editors are expected to include a brief
summary of the issue in subsequent working drafts until the
issue is resolved. The people who raised the issue are
encouraged to produce the wording for such notes.
It seems to me that first the issue must appear in the issue
tracker. Merely to get things into that tracker etc has until now
been a complicated task.
Keep in mind that Sam said it has to be "3 indpendent members" who
request that it is added as a note in the spec. And I suppose that
he judge "indpendent" in the same way in this context as else.
Also, the burdon on finding the text for that note, is on the
requesters.
Thus there are 2 steps to pass for a given issue, before this can
happen.
I see value in recording it in the spec. And I would not make the
comparison with software engineering too literal. The main product
of this group is a text. And keeping things in the text could
contribute to the "testing of the application" - namely the text.
Text bugs may fit directly in the text.
regards,
Leif Halvard Silli
