[ On Sunday, February 13, 2000 at 22:30:52 (-0700), Tom Satter wrote: ]
> Subject: Re: CVS File Locking
>
> What I was pointing out about the 'no compromise' position
> that Greg has taken is that in the CVS add discussion,
> almost everyone has agreed with having code to support the
> new syntax (part one of the paragraph above). However,
> there has been serious questioning of the value and
> complexity of implementing the code for looking upward
> in the repository for working directories. Greg's response
> to all of these posts has been of this type (quote):
>
> > I'm going to simply ignore the counter-proposal you made -- it has no
> > meaning at all to me until someone can show how looking in the parent
> > directories can somehow cause harm under normal circumstances. I'm
> > going to continue operating under the assumption that no harm can be
> > done and that abnormal circumstances can be detected and/or documented.
>
> I think that we would all agree that you could implement
> almost anything into CVS using the argument that 'I am
> going to ignore you until you show me how it could cause
> harm.' Heck, with the additional phrase on the end about
> detecting abnormal circumstances we could probably even
> put an MP3 player into CVS without causing any harm.
I was going to not reply to this, but I think I can't leave such a
horrendously nasty mis-representation of my argument stand.
You have taken my words completely out of context.
Look, there have be *NO*, none, zero logical explanations of why it
could be dangerous to look back down the chain of parent directories for
the existence of a valid working directory. Until someone can show me
even one I cannot treat any "questioning" (seriously intended or not) of
my proposal as serious. There is no logic in adhering to a purely
emotional "worry" about doing something that some people seem to
perceive as somehow dangerous, especially when all level-headed
investigation and logic points completely the other way. There isn't
even anything radical about looking in the parent directories! If I
hadn't had the same irrational reaction on first sight I'd be totally
stunned by the nonsense people have been worrying about on this issue.
Luckily I quickly got over my worries after I looked at that facts.
Next time you discuss a pet idea with your friends and then some goof
comes along and tries to tell you that you *MUST* toss it away and do
something completely different and justifies his demands with totally
illogical statements how are you going to respond?
I most certainly will not compromise my ideas just because some people
say completely illogical things and make completely unreasonable demands
that cannot be justified in any way. If anyone thinks I'm going to
change my plan for them even though they can't provide one shred of
evidence that I'm wrong or in danger then they can just go fly a kite
(and hopefully they will do so and thus leave me alone!).
I'm just sorry that I've wasted a whole lot of my time trying
unsuccessfully to clarify a few simple concepts and trying to correct a
whole lot of really broken perceptions when I could have been coding.
> The point that has been made over and over is that
> looking upward in the directory structure complicates
> the way that CVS works.
No, it does not. Those who have made such claims are completely wrong.
The minor increase in complexity comes in making CVS do all of the other
parts of the proposal. There is almost no increase in complexity due to
having to look back at parent directories to discover if we are within a
workspace and to populate the intermediate directories with
administrative files as necessary. Don't believe everything some of the
other noise makers are saying. You don't have to believe me either --
why don't you try implementing my proposal sans the parent peeking?
> It makes the add command be
> different from all of the other commands
No, it does not. You are completely wrong.
How many million times must I say this? RTFM! The user interface of
the current implementation of "cvs add" is entirely different from any
other sub-command that operates within a CVS managed workspace. I have
simply proposed bringing "cvs add" into line with all these other
commands by making it treat its command-line arguments in exactly the
same way that the rest do. RTFM again.
> and adds in
> complexity that CVS does not NEED in order to be a
> great concurrent revisioning system.
In fact CVS does need the new ability I've proposed for "cvs add" -- it
has been a frequently requested feature ever since the very beginning of
its public release.
My proposal kills at least three major long-standing problems with one
relatively simple and elegant solution.
--
Greg A. Woods
+1 416 218-0098 VE3TCP <[EMAIL PROTECTED]> <robohack!woods>
Planix, Inc. <[EMAIL PROTECTED]>; Secrets of the Weird <[EMAIL PROTECTED]>
