Re: Delphi Bugs
From: Bob Dawson (RBDawson_at_prodigy.net)
Date: 07/31/04
- Next message: Xavier Pacheco: "Re: Ideas for my site/workbook"
- Previous message: Andrea Raimondi: "Re: Ideas for my site/workbook"
- In reply to: Derek Davidson: "Re: Delphi Bugs"
- Next in thread: Derek Davidson: "Re: Delphi Bugs"
- Reply: Derek Davidson: "Re: Delphi Bugs"
- Reply: ozbear: "Re: Delphi Bugs"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Sat, 31 Jul 2004 10:55:06 -0500
"Derek Davidson" wrote
>
> Why is that relevant? As you rightly state, all code is owned and
> no self-respecting professional owner is going to allow untested
> code into a tip revision. That would be dumb.
Read: There's no such thing as an open source method for major projects; the
bizarre is a myth.
> But there's nothing stopping ANYONE from changing their copy
> of the code, making use of it themselves
Nor is there any advantage to it. Without a full owner's perspective,
personally branching someone else's code is simple folly, for multiple
reasons stated previously. That's just not something a professional
developer has time to do.
> and uploading it to the owner for them to review, test and (hopefully)
> incorporate in the next code revision. You CANNOT do this with
> closed source.
Au contraire.
Anyone who finds a bug in Delphi, JBuilder, etc. can document it, submit it
to Quality Central for Borland to (using your exact words) "review, test and
(hopefully) incorporate in the next code revision." The only difference is
whether the submission is in the form of provable bad behavior (i.e. a
reproducible test case) or that PLUS a proposed code branch.
For the largest and most significant kinds of bugs--logical errors, missing
logic, design errors, design inconsistencies, unrealized
assumptions--submitting reports at the provable behavior level is orders of
magnitude easier than determining the fix, since the fix may require code
changes in multiple locations, may impact on other code, may impact on other
planned changes, etc. It's simply silly for anyone _not_ the code owner to
try to specify changes at that level. I'll argue with Danny, John, Allen, or
any other Borlander about behavior expectations, but I don't have the hubris
to pretend I'm going to write their code for them.
Or to go back to the 'topic case'--if Danny had rewritten the way Linux
handles library loads, would Linus have accepted it? Doesn't sound too
likely, as Linus apparently believes the behavior 'correct' for historical
reasons. Same at any level--in my work, if a user passed me a code change,
I'd ask him what behavior he was really trying to change. I would be
extremely unlikely to have any interest in looking at his 'fix' until I
understood that. Owners change code--that's what being the owner means.
The only case in which a code-level submission would really be meaningful is
that of a simply and demonstrable code error, such as
for i := 0 to Something.Count //sic Count -1
But those aren't particularly hard bugs to find in the first place, and even
they may raise regression test issues.
> buy so we can be sure to have control. I suspect NONE of us buy
> without because we can "hack it and do it anyway".
There's a considerable difference in complexity between a component and an
OS or development environment. And even there I think your argument fails.
Yes, I'm licensed for the QuantumGrid with source--but I'm not likely to try
to fork it. If I have a problem with behavior I'm not going to try and
rewrite it, I'm just going to report the behavior; fixing it is DevEx's job.
If the problem is a show stopper, then I look for another vender--it's that
simple. We vault the source code as a versioning issue, not because we have
any expectation or interest in working on it.
> have shown (and with which I agree) is that all code is owned.
> I'm as surprised by that as I was when I discovered that the
> Pope was a catholic.
Still is as far as I know <g>. Yes--all code is owned, and thus all code is
the product of a cathedral approach--don't see why that implication is so
hard to follow. Which does mean that source code is effectively read-only in
public places.
As for personal branches, telling someone that they'd be better off with
open source presumes they have the time, background, and need to care.
That's an extremely rare conjunction. In the real world, if Delphi was open
source and I found one of the developers who works for me patching the
compiler or rewriting the IDE, I'd counsel him pretty plainly about 'career
limiting decisions.' We've got our own work to do.
bobD
- Next message: Xavier Pacheco: "Re: Ideas for my site/workbook"
- Previous message: Andrea Raimondi: "Re: Ideas for my site/workbook"
- In reply to: Derek Davidson: "Re: Delphi Bugs"
- Next in thread: Derek Davidson: "Re: Delphi Bugs"
- Reply: Derek Davidson: "Re: Delphi Bugs"
- Reply: ozbear: "Re: Delphi Bugs"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
