[Ffmpeg-devel] SVN dump

[Måns Rullgård]

Attila Kinali wrote:
> On Wed, 18 Apr 2007 10:06:44 -0400
> Dave Dodge <dododge at dododge.net> wrote:
>
>> Do you have Java support on your target system?  Jython supposedly
>> implements Python in pure Java (it compiles Python to JVM bytecode).
>> I don't know if Mercurial would easily run on top of it, though.
>
> I hope you have your fire proofe underware on. Suggesting Rich to
> use Java is even worse then suggesting to use Python. Now to use
> Java to run Python scripts is a whole class of its own.

I have to agree with Rich's supposed opinion here.  It is beyond mad.

>> > And writing your program in a RAD tool/prototyping/script language
>> > does not speak well for its quality.
>>
>> Not sure I entirely agree with that.  For example if the program
>> needed to do a _lot_ of string processing, it would probably be easier
>> and clearer to use some higher-level language than plain C.
>
> Well, i don't agree with Rich on that, but neither do I with you.
> But that's beyond the topic here.

I don't know why, but I tend to disagree with Python aficionados on
more than mere choice of language.  Also, reading Python code always
makes me feel uneasy.

>> Aside: it could be worse; one of the main "competitors" to Mercurial
>> and GIT is Darcs.
>
> Well, i tried darcs about 3 years ago. Set up a test repo, did
> a few things i was used to do with cvs and run into problems
> that darcs couldn't solve. It was to some extend a bug of darcs
> (it was supposed to be possible), but a look at the code revealed
> that it was a design problem rather than just a bug. Questions
> about this on the ml were simply ignored.
>
> No idea whether they fixed these issues by now, but then darcs
> pretty much died for me.

Out of interest, which were the operations that failed?

>> While it has the best (or at least the most
>> interesting) design theory behind its operation, it ends up being slow
>> for large projects, apparently has some corner cases where it goes
>> into a near-infinite loop, and is implemented in Haskell -- making its
>> code mostly impenetrable to the average programmer who might be
>> interested in working on it.
>
> Haskell is only difficult if you cannot think in a functional
> language. Ok, maybe Haskell is not the right one to start,
> try maybe Scheme first (has one of the smallest language
> descriptions known to me). And IMHO every good coder should have
> at least once tryed a functional language.

Agreed.  To quote esr:

LISP is worth learning for a different reason [than Perl] ? the profound
enlightenment experience you will have when you finally get it.  That
experience will make you a better programmer for the rest of your days,
even if you never actually use LISP itself a lot.

That said, the more elegant the language, the nastier the implementation
tends to be.  This suggests that many of these languages are best left
to the academics and not used for real-world projects.

-- 
M?ns Rullg?rd
mans at mansr.com