On Thu, 2009-03-19 at 08:17 +0100, Torsten Bronger wrote: > Hallöchen! > > Malcolm Tredinnick writes: > > > [...] > > > > Even PHP: I mean, Flickr, Wikipedia, Yahoo -- these are some > > pretty large sites running on PHP. > > I've always wondered why anybody uses something non-Java for Web > applications. Given that Java is faster than PHP, Python etc., this > also means that you need less computing power in your server farm. > On the long run, this should *always* be worth it financially, > unless Java is a nightmare to maintain, which I don't assume.
There are a lot of assumptions in that a paragraph which may or may not be true (particularly the last part depends on a lot of variables and I would default to not-true unless the developers have been super-careful). My professional experience is certainly that Python is a lot easier to maintain than the equivalent program in Java (I've actually had the good or bad fortune to help develop more or less exactly the same program in Java and Python, so I have at least one medium-sized, multi-machine project to use as a comparison). I don't think I've ever been swayed by any blinding speed difference of Java over Python on very many different projects. You can certainly find projects where Java or C will be a faster choice. I can also find code where Python wins unless you put a lot of effort into ironing out problems (doing work with lots of string manipulations is almost the exact opposite of fun in Java, for example). > Granted, I use Python for our institute's internal Web application, > but I expect our traffic to be smaller than what one single > processor can handle. > > Is it because the network and the database are the limiting factors? > But even then, the part of your server farm running the interpreters > could be reduced. The difference isn't going to be *that* significant when you measure it, in a lot of cases, I suspect. You're right that the database and network are often limiting factors. Although, we (large system developers) also spend a lot of time removing the need to do things like database accesses, or even disk reads. At that point, all the languages are converging on things like the speed of your caches and communication speeds. For web applications, you're hopefully not doing that much computation. If you are, you're passing it off to native system libraries. There are also considerations such as Java using a lot more memory in cases than an equivalent algorithm in Python. On the reduced server farm load, remember that things are discrete. Saving 0.05 of the server load doesn't mean you now only have to buy 0.95 of a server. You still have to buy them in whole units. When combined with the natural wish for a bit of overhead on busy sites, and you might end up not saving that much (for which you've traded off things such as, possibly, maintenance and development speed). > On the other hand, PHP is *highly* popular. So why is this? You realise you've now guaranteed this thread will go on for about 50 messages, right? :-) My take: PHP isn't a horrible language to program in. It's got some flaws, but if you really understand the language it's easy enough to work around them. A lot of its bad reputation comes from software that *uses* the language and does things badly. The language itself isn't necessarily bad. I would say it makes it a bit easier to shoot yourself in the foot in the web space than Java or C or Perl, but Python has some cool ways to do that, too (I'll argue that taint mode in Perl makes it a slightly better native web language than Python in that respect). The barrier to initial entry for PHP coding is very low. This means, of course, that the number of beginning coders in that space is very large and many of them don't move up the learning curve. So popularity by raw numbers favours the simpler languages (BASIC was also very popular when I was a young 'un for the same reason; then again, so was 6502 assembly, so my impressions may not be too relevant these days). Popularity on the right-hand side of the curve (call it "expert level" or "large, professional systems") is interesting and I don't have the information to make a judgement. Except for this: the fact that Yahoo have moved to PHP internally (they actually moved from "other stuff" a few years back) and things like Flickr are built on it (independently, prior to being bought by Yahoo, presumably) means it has something going for it. I will note, though, that PHP has some of the best integration with a web browser of any language (including Java). Have a read of Ian Bicking's piece comparing mod_php to mod_python (et al), for example. He gives a good description of that strength of PHP as a language as an implementation for Apache. I would love for somebody like Cal Henderson to make their next book be "Writing enterprise systems in PHP", except without the "enterprise" word, because I suspect there's a lot to learn there. Again, raw language popularity might have helped drive the adoption initially and it's turned out to hold up. I've been involved in Python programming for paid work for a bit over 10 years now and, until very recently, have experienced the same problems Shabda is trying to mitigate: corporate reluctance to use this odd language. Sometimes it's used and sometimes it isn't. Ditto with Perl, PHP, Ruby... For many problems, I'll precis what I wrote in the original mail: there isn't a huge difference between the functionality available so it often comes down to the quality of the talent you have available. Either in house or for-hire. End of the day: pick any reasonable language and I'll wager good programmers can developer a product that performs adequately for almost any task. Sure, some languages will be better than others for some problems. But in the fully general case, I'm happy saying it's a bit of a wash and people should use whatever they feel most comfortable with. Regards, Malcolm --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to django-users+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---
