On-Rev founders - some speed data?
benr_mc at cogapp.com
Fri Apr 24 06:47:32 EDT 2009
On-Rev founders are in a unique position at this point to give some indication
of the performance of the new facility (granting of course that there may be
also sorts of debugging etc going on, or at least further optimisations
possible in the future).
I don't think that speed is going to be the main issue in choosing a
server-side scripting language (and I know that it's rare that you can't make
more of a difference by improving the algorithm than by changing the language)
- but I think that perceived or presumed lack of speed could be an argument
against a new entrant in this market.
My experience with Revolution on the desktop is that I've recoded C apps into
Revolution and found that my app got faster - because I was standing on the
shoulders of Scott Raney, Mark W et al who'd put serious efforts into
optimising facilities. Of course there are also odd things that are very slow
(you can crop an image faster in Transcript than using the built-in command,
whch could perhaps be taken as a tribute to the speed of all the commands
other than 'crop') - but in general my experience is that Rev apps are not
only fast, but faster than non-Rev programmers assume. They think that we get
fast development time at the expense of slow execution time - I think we often
get both. But against scripting languages on the server we're in a different
realm, and I don't know what we should expect.
Has any one done anything like this? Ideally, someone might have taken an
existing facility in PHP or Python or Perl or Ruby (is Ruby installed on
On-Rev?) and recoded it in Rev. Otherewise googling eg "benchmark php python"
produces a lot of less-than-real-world tests that have been translated into
various languages for comparison. I'd be very interested to know how on-rev
It would be great if someone who has access to the new kit could do some tests
which for the first time can be running on the same server in the same
conditions (as Apache module rather than CGI) and hence give some kind of
actual data points.
More information about the use-livecode