Erik Anderson wrote:
>I'm going to hazard a flamewar here and interject that MS's concept of
>"shelveset" is not necessarily BS even if it can be done with SVN.
>
>I'm seeing that the difference between a lot of what MS does seems to be
>the difference between concept and implementation. SVN is a well thought
>out concept and can be made to do most everything that people need in a
>versioning system, including these "shelves". It is rock-stable and does
>its job well, with many configuration activities pushed off to the person
>who sets up the repository.
>
>
Right so far...
>However, MS appears to be less interested in concept (which it can get
>from existing projects) and more interested in implementation. That is,
>this is a powerful system, but can it do what the user wants to do without
>requiring a lot of hullabaloo involving shell scripts or strange menu
>systems. The TortoiseSVN project is heavily invested in improving the
>"implementation".
>
>A lot of the talk on this list the last few weeks (that I've been here)
>has been discussing this balance of concept vs. implementation (the "meta
>tags" discussion is an excellent example of this).
>
>I came up with this idea after seeing all the junk surrounding the Java /
>.NET "war". One of the biggest selling points I was hearing was that
>".NET has a JIT compiler and Java does not". Besides being wrong, this
>highlighted the idea that Sun was more interested in getting a strong
>language that could do the job well while offloading tools like optimizers
>and JIT compilers to be done by interested third parties (like TYA)
>
>I would like to avoid the whole Java / .NET flamewar here if possible, but
>the concept I am trying to put forth is that both the core subversion and
>the ease of user accessibility may be important in the future of this
>product.
>
>
You're right as far as it goes. There's one problem with your
comparison, though: there's a serious difference in, let's say,
deployment scope. MS's TS (and to some extent, TortoiseSVN) serve a much
smaller set of configurations than Subversion in general. MS usually
tells people to use their product in a certain way, and you don't have
much leeway. Subversion does exactly the opposite.
This doesn't mean that improving usability is not an interesting goal
for Subversion; but it does mean that we have to be very careful about
including usability-oriented features. Those that we accept must be
useful for most users, not just a small group (even if that group sees
their way of doing things as "obviously correct" :)
What we _do_ provide is enough flexibility in the API to allow people to
put the usability features they want in a customized client, of which
TortoiseSVN is a first-class example.
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Aug 3 01:10:26 2005