Subject:
|
"the new way" (was: Re: Self-Certification in parts tracker)
|
Newsgroups:
|
lugnet.cad.dev
|
Date:
|
Wed, 12 Jun 2002 15:43:39 GMT
|
Viewed:
|
406 times
|
| |
|
|
In lugnet.cad.dev, Franklin W. Cain wrote:
> Personally, I do *NOT* like the new way of doing things,
> because it's just too slow.
Actually, we're on track this year to release approximately 800 files.
That's as many files as we've ever released, back in 1998 and 1999.
Here are the released file counts, starting from the first 'LCAD
Update'.
Year Files
1997 128
1998 806
1999 685
2000 376
2001 224
2002 361
> I *much* prefer the timeliness
> of the original way of getting files released, where *everyone*
> was expected to review *every* file being considered for release,
> and vote accordingly.
Except that not everyone looked at everything, and frequently stuff got
through that shouldn't have, and it was a big load of work for one
person (that last part is the most significant; one person doesn't scale
well). Most files were 'rejected' because the admin caught them when
they were submitted, and sent them back for corrections.
The old system just did not work. And it took too much time from one
person. I *really* don't understand how Terry managed to put out so
many updates those first years, considering that he was doing everything
manually. I managed to automate a lot of the work, but it was still a
very large pita.
> But, they wanted to change things,
> to install better safeguards, to ensure a higher standard
> of quality for the files to be released. I certainly don't
> begrudge "quality", but I cannot believe that we cannot
> streamline this new process to make things, if not quite
> as fast as originally, at least not as SLOW as things have
> become...
I'm open to suggestions. Seriously. One thing I can think of right
away:
Start an 'issues list' for parts, where defects can be recorded, along
with the current status (open, confirmed, fixed, dropped, etc).
Reviewers could use this list in two ways. First, it could influence
their decision on whether to certify or hold files. Second, they could
record issues they discover, without having to hold up the release of a
file. Of course, authors would also refer to the list, to track what
needed fixing. And anyone looking for something to work on could cruise
the issues list, and address any open problems.
> And in my defense, let me point out that I can think of at least
> two files of mine that were *this close* to being released
> (one was "wtg. admin", the other was "cert.") that I re-submitted
> (which WIPED OUT ALL "CERT." VOTES), just because I had discovered
> flaws in them.
Let me suggest that this affected your perceptions more than it affected
what actually happened. Imagine that both files became certified and
were released, and *then* you discovered the flaws. What would happen?
You'd fix the files, send them to me, I'd resubmit them, and they'd be
in the same position (on the tracker, no votes) they ended up in anyway.
The differences would be:
- *Some* version of the file was released.
- We put in more work on the files overall.
What's the alternative? Should votes stay as-is when a new version is
submitted? Two problems I could see with that:
1. Getting the hold-reviewer to go back and re-review the file might be
a problem -- people leave the team, don't have interest in the part
anymore, are stubborn, etc.
2. The 'new version' could be a complete re-write, and *should* be
totally reevaluated.
Would it help if the detail page provide datdiff links, to compare older
versions to the current version? Would that make it easier for people
to re-review files that are updated?
Steve
|
|
Message has 1 Reply:
Message is in Reply To:
 | | Re: Self-Certification in parts tracker
|
| (...) I do *NOT* recall this discussion. Was this on LUGNet? How long ago? Or, was it on "RTL"? (Does anyone still use "RTL"?) Also, FWIW, I asked Steve, before I ever did any self-certification, if he had any qualms about this. He said that if (...) (22 years ago, 11-Jun-02, to lugnet.cad.dev)
|
18 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
