[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [lord@regexps.com: business models and revision control]

From: Tom Lord <lord_at_regexps.com>
Date: 2002-08-13 04:20:19 CEST

> Perhaps you could clearly and briefly summarize the point
> that you're trying to make

I think the design and implementation of svn is broken in some serious
ways. I _might be_ mistaken, however, I formed this opinion late last
year and the further I've progressed with arch, the more I've become
convinced. If I am right that svn has serious flaws, then I also
believe that a 1.0 deployment of svn can do serious harm to the

Faced with that, and _not being a svn expert_, I've been trying to
raise the technical issues with you since at least january in ways
that do not _presume_ svn is, in fact, broken. More recently, I've
been compelled by circumstance to make a stink about the practical,
political, and economic issues that I think are related to the lack of
uptake on these issues from the svn developers, and my inability
basically to raise money for an alternative approach.

I read replies like (paraphrasing) "awk is icky" (true enough in the
general case -- not obviously relevent to this specific application)
and "maybe that'd be nice to possibly talk about after 1.0" as
operationally equivalent to "talk to the hand". Basically, I keep
having to surpress my rage at what seems to me like your incredible
rudeness -- a condition I'm resigned to believing is probably mutual.

Here's what I told one old friend recently:

    My basic take on svn is that it has some good ideas and ok coding
    practices but a flawed design, perhaps from the ground up, or
    (optimistically?) in the details.
    To put it pessimistically: it's (at least):
            1) too complicated
            2) depends on too many other pieces
            3) too centralized
            4) peculiar (at best) user interface
            5) questionable storage manager semantics
            6) out of control (judging by bug reports on the dev list)
    I'm sure I could go on.
    A FS with cheap tree cloning is a good thing. The rest of it is just,
    based on everything i've ever learned, pretty much wrong.
    Again -- that's the *pessimistic* take.
    Optimistically, maybe it's:
            1) more modular than it first appears
            2) with a storage manager that can be easily fixed and,
            3) that hits an interesting sweet spot,

Now, to be clear, arch is very far from perfect itself. It's a total
rush job. I said your coding practices are just "ok" -- well, the
practices behind arch are "rush job" with all of the attendent
problems (there exist better examples of my _patient_ work).

Getting any clearer? I fear this just comes across as rude or
attempted-thunder-stealing. Really, all those attempts to start
threads about patch set formats were the best solution I thought of to
avoid those appearences. I was attempting a kind of socratric
leading-question type dialog (but do _not_ read into this an attempted
teacher/pupil thing -- i just meant that seemed like the least
ego-involving mode of conversation to start with.)



To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Aug 13 04:13:01 2002

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.