[python]
[1/3] from: maximo::meteorstudios::com at: 18-Nov-2003 13:57
I thought we had a hard time getting concensus...
boy I've been on the python list for a day and got 150 mails.. many of them are down
right rude.
python may have more users, but its also a wild jungle.
its also found it fun to follow a thread called:
Re: Too much builtins (was Re: Python's simplicity philosophy)
when I know that it takes 50 modules to do what rebol does with the amount of code in
the equivalent of 3 of them (without needing to actually load them ... ;-)
This while running 5-10 times faster... and having a much more reusable code base in
any circumstance.
They still have nice souls on the list, I got an answer within the day like here.
But when reading about the more "philosophical" debates, like we sometimes have, there
seems to less intent on cooperation and solutions than on proving someone's point.
no need to follow up, it was just a short message to illustrate how this list fares towards
its "competition" :-)
-MAx
---
You can either be part of the problem or part of the solution, but in the end, being
part of the problem is much more fun.
[2/3] from: tim:johnsons-web at: 19-Nov-2003 17:17
* Maxim Olivier-Adlhoch <[maximo--meteorstudios--com]> [031118 10:20]:
> I thought we had a hard time getting concensus...
>
> boy I've been on the python list for a day and got 150 mails.. many of them are down
right rude.
I program about equally in python and rebol....
> python may have more users, but its also a wild jungle.
>
> its also found it fun to follow a thread called:
>
> Re: Too much builtins (was Re: Python's simplicity philosophy)
> when I know that it takes 50 modules to do what rebol does with the
> amount of code in the equivalent of 3 of them (without needing to
> actually load them ... ;-)
I am glad that the python community is addressing this issue - I just
spent the better part of two days trying to resolve a pyGTK
interface, Yikes! what a can of worms....
> This while running 5-10 times faster... and having a much more reusable code base in
any circumstance.
I've really not seen such an all-around speedup of rebol over python
to that degree. What comparisons have you made?
I would feel safe suggesting that line-for-line, (speaking from my
experience only) rebol is probably 50% more productive than python.
Python appears to take cues from other languages with their adaptation
of list comprehensions. It would be nice if they could blur or
eliminate distinctions between assignment and code like rebol does.
> They still have nice souls on the list, I got an answer within the day like here.
Both MLs are very helpful, yes I have found that too...
> But when reading about the more "philosophical" debates, like we
> sometimes have, there seems to less intent on cooperation and
> solutions than on proving someone's point.
I do find more opinionated positions on the python ML, but then
1)python is more 'established' than rebol. (older 'community)
2)Python implements a more 'conservative' or 'structured' or
'disciplined' (pick your adjective here) approach than rebol.
> no need to follow up, it was just a short message to illustrate how this list fares
towards its "competition" :-)
I tend to make the opinion here from time to time that rebol needs to
build up a distribution package like python and perl have. That would
attract more programmers to it (IMHO), and that would include my
partner who is a dedicated 'perlhead' mostly because of all the
modules.
On the other hand, care must be taken to insure that 'resources' don't
become too dependendant-heavy, which probably lead to the python
thread that you're referring to above. And my nightmarish experience
that I refer to above.
MTCW:
tim
--
Tim Johnson <[tim--johnsons-web--com]>
http://www.alaska-internet-solutions.com
[3/3] from: moliad:aei:ca at: 23-Nov-2003 12:16
---Tim Wrote:---
> * Maxim Olivier-Adlhoch <[maximo--meteorstudios--com]> [031118 10:20]:
> >
<<quoted lines omitted: 8>>
> experience only) rebol is probably 50% more productive than python.
> [...]
I was basing that on a mail posted on this list (sometime ago) from someone who had done
in-depth benchmarking.
it is possible that the gap between the two has lessened since then, especially since
python now uses more and more C based compiled object modules, which, like any rebol
native, execute much faster.
you are right, in what I can see from more recent comparisons. they currently seem pretty
equal in speed.
one thing though is that when writing code in rebol, performance is very elastic. yesterday
I was having fun profiling a hashing loop:
for 100000 ops, time went from:
50.1 seconds
to
4.02 seconds
A 92% time reduction, changing the block type to hash! lowered time by 90% (down to 5
senconds). some details like the way items where added to the block (insert instead of
append), and using repeat instead of 'FOR trimmed about 20% in remaining time, from 5
seconds to 4 seconds.
in other languages, there is usually less elasticity because there is many times only
one way to address such low level things as setting/finding a value in an array... in
rebol there can be many compatible approaches.
-MAx
Notes
- Quoted lines have been omitted from some messages.
View the message alone to see the lines that have been omitted