PHP functions originally bucketed by strlen, were renamed to balance length

http://news.php.net/php.internals/70691

by mappu • 13 years ago

140 156 13 years ago

news.php.net

mistercow5• 13 years ago

Sometimes I feel like Rasmus Lerdorf is going to come clean one bright April Fools' Day, and reveal that PHP was just a hilarious joke that sort of got out of hand.

jtreminio4• 13 years ago

Yeah! Haha the joke's on Facebook, Wikipedia, Flickr, Yahoo! Answers, Photobucket, Digg (RIP), etc ... Those fools have been using a joke language and didn't even know it!

Seriously, the PHP hate gets old. It started off very much as a scratch to one man's itch. That man wasn't the best programmer ever (wouldn't even say top 10%, or even close), but he had an idea and executed it to make one of the most successful free products ever that has generated billions and billions of dollars for the world economy.

Can you say the same?

cmelbye2• 13 years ago

Does the fact that it's used by large websites change the fact that it's a poorly designed language?

Iftheshoefits2• 13 years ago

Arguments from popularity can be amusing when laden with sarcasm, but that doesn't make them valid arguments.

I'd say PHP is and has been popular more because it's easy to hire an army of self-trained "php developers" for peanuts than because it is a "good" language.

kailuowang• 13 years ago

I don't know if you can say that "those fools have been using it and don't even know it." It's hard to tell if they had a time machine would they go change that decision (to use php). Anyway I agree that there has already been too much time spent on this language, no need to waste more beating the dead horse.

mehwoot• 13 years ago

that has generated billions and billions of dollars for the world economy.

That's an interesting thought. Has it? I wonder what a lot of websites would have used if PHP has not been around. I wonder if, overall, PHP existing has actually generated billions and billions of dollars in value.

jfb• 13 years ago

Who cares? It doesn't make the language any better.

eevee• 13 years ago

Conversely, I wonder how much time/money has been spent porting away from PHP, or working on stuff like hiphop?

I don't have an estimate, but it's an interesting question if you're going to argue in terms of economic impact.

mistercow5• 13 years ago

I'm not really making any strong comment about PHP's usefulness or anything. I'm just saying that many of the things that have happened in its development would sound like they came right out of the manual for INTERCAL, except that they're too over the top.

As for its popularity, well, that would be the "got out of hand" part.

MichaelGG6• 13 years ago

Joke is sort of on them. Every single article I've read with engineers at the big sites talking about PHP mainly deals how to get around PHP. All of the things PHP is lauded for making things "easy" are the exact things the big sites go out of their way to avoid.

Those sites are popular in spite of PHP, not because of it. If Mark had been proficient in ASP instead of PHP, FB would probably be running .NET. It's not like anything early FB did was particularly impressive and relied on special language features. (Correct me if I'm wrong.)

freshhawk• 13 years ago

Crack cocaine is a very successful product that has generated billions and billions of dollars for the world economy.

You have to watch out for arguments from popularity, they don't prove or even suggest anything other than popularity.

themodelplumber• 13 years ago

https://wiki.theory.org/YourLanguageSucks

gst• 13 years ago

I like his quotes:

I'm not a real programmer. I throw together things until it works then I move on. The real programmers will say "yeah it works but you're leaking memory everywhere. Perhaps we should fix that." I'll just restart apache every 10 requests.
I was really, really bad at writing parsers. I still am really bad at writing parsers. We have things like protected properties. We have abstract methods. We have all this stuff that your computer science teacher told you you should be using. I don't care about this crap at all.

(from http://axonflux.com/5-quotes-by-the-creator-of-php-rasmus-le...)

GhotiFish5• 13 years ago

In a very strange way, I respect this man.

nivla• 13 years ago

I find comments like these interesting and at the same time ironic, especially on a sites like HN. I could think of 10 reasons why PHP is the "wrong" language but I could also think of 100 reasons why it made a successful one. Rasmus my not have created a perfect language but he sure did create one heck of a successful one which even the "elitest"/"perfect" language are still trying to catch upto.

Never belittle other's work, especially when striving to discover that magic formula to success. All the posts on HN about "How we made X successful" and we brush off the obvious ones in front of our eyes? Ever stopped to think why a crappy X is popular? Maybe there is something to it.

Also not to mount on irony but keep in mind, you are commenting on a site that has the HTML syntax from the Geocities era (it has its reasons though).

GhotiFish5• 13 years ago

I would be very interested to hear those reasons.

MichaelGG6• 13 years ago

Can you explain how, in any possible universe, using strlen as a hash function and choosing function named to bin on length in any way helps PHP, at all?

skloubkov3• 13 years ago

While I understand why this was done (for the sake of comparability), I feel like its better to bite a bullet and slowly phase out broken functions over several versions.

Warn initially (few versions) and then raise exception/critical errors after. If developer isn't willing to update, chances are server isnt being updated either anyways.

jtreminio4• 13 years ago

Nothing in this discussion mentions anything broken, though.

Yver2• 13 years ago

Indeed, and the title is incorrect too. Function names were not renamed to end up different buckets; Their names were picked that way.

Someone• 13 years ago

What about usability? "I wasn't too worried about not being able to remember the few function names." certainly indicates that usability decreased. That may have set a pattern for it getting broken, eventually.

damncabbage• 13 years ago

Well, not really. Consider the (very common) case of a PHP-based blog, forums or shop being hosted on shared hosting. The latter gets an upgrades, and the site explodes.

(I say do it anyway. It's like IE6; give it a long grace period, then tell people they need to get with the program.)

shoyer2• 13 years ago

Better idea: bite the bullet, drop PHP. Switch to a language that was designed. In the eyes of many of us, PHP is beyond redemption.

skloubkov3• 13 years ago

Difficult to get as sign-off on re-write of large systems when there technically nothing broken.

Killswitch6• 13 years ago

Better idea: Be a better programmer and don't blame a language because you suck.

mikeash2• 13 years ago

The site is extremely slow at the moment. In case anyone has trouble accessing it, I managed to load it, and here's the relevant bit:

    Well, there were other factors in play there. htmlspecialchars was a
    very early function. Back when PHP had less than 100 functions and the
    function hashing mechanism was strlen(). In order to get a nice hash
    distribution of function names across the various function name lengths
    names were picked specifically to make them fit into a specific length
    bucket. This was circa late 1994 when PHP was a tool just for my own
    personal use and I wasn't too worried about not being able to remember
    the few function names.

tptacek2• 13 years ago

That can't possibly be true. I refuse to believe it.

pygy_• 13 years ago

Either way, it's hillarious :-)

zaphar• 13 years ago

Oh It makes complete and total sense. PHP was not a "language" originally. It was a collection of handy functions in Rasmus' toolbox. When it "graduated" to language, and I use that term loosely no one bothered to clean it up. No one bothered to clean it up is PHP's mantra and it explains everything that people hate about the language perfectly.

mathrawka• 13 years ago

Well, good thing that PHP is open source and we can just pop a visit to http://museum.php.net/php2/ to actually see that it is true.

M4v3R• 13 years ago

Taken from PHP 1.99 source [1]: https://dl.dropboxusercontent.com/s/e8k4n2mx1y5vsa4/Zrzut%20...

So yeah, it's true :).

[1] http://museum.php.net/php2/php-1.99s.tar.gz

kingkilr• 13 years ago

It's a joke. A clever satire. I simply refuse to believe it's real.

jMyles• 13 years ago

Yeah, well, I for one just spit beer out of my mouth at a fucking public sports bar.

TOGoS• 13 years ago

That he futzed around with function name lengths instead of just fixing the hash function is a pretty good analogy for both the design of the language as a whole and to much of the code that's written in it.

Queue "it's popular and therefore must be good" trolls.

jarek2• 13 years ago

How about spelling trolls? It's "cue":

"From the letter Q, abbreviation of Latin quando (“when”), marked on actor's play copy where they were to begin." - http://en.wiktionary.org/wiki/cue

untog• 13 years ago

"it's popular and therefore must be good"

Feels strawman-ish, I've never heard anyone say that. "It's popular therefore it is a legitimate language choice", sure.

jmccree3• 13 years ago

PHP was doing MVP before it was cool.

MichaelGG6• 13 years ago

I understand it but don't get it. Why use strlen? Even a simple one-line character-xor does better. (I tested using the 354 function names from the PHP2 source code.) Why would you bother bucketing, but then take such a silly approach and actually aim to name things for length?

dscrd• 13 years ago

Perhaps, and I am being totally serious here, Rasmus did not know about xor.

Tloewald• 13 years ago

Well, points for honesty, I guess.

kabdib• 13 years ago

I've written a few thousand lines of PHP in the last year or so. Every time I do, I want to weep.

It might be better than Perl, but not by much.

smacktoward2• 13 years ago

Like Perl, PHP is a long-lived, heavily-used-in-production language, which means there is a lot of cruft that has accumulated over the years that's difficult for the platform developers to just slough off. But much like there is a Modern Perl movement that has coalesced around voluntarily cutting out cruft and using a much cleaner, nicer-to-work-width subset of Perl, there is a similar movement in PHP-land that's doing the same thing.

PHP The Right Way (http://www.phptherightway.com/) is a good place to start if you're curious what "modern PHP" looks like.

forsaken• 13 years ago

Those same arguments would apply to Python too, and it doesn't seem to have nearly as much cruft.

Killswitch6• 13 years ago

This. Nobody taken seriously in PHP world doesn't use modern PHP tools, Good frameworks, Composer, follow PHP-FIG standards, all that stuff... Just because you had to maintain a PHP4 app 10 years ago, doesn't mean PHP hasn't changed. And it also doesn't mean real devs don't have standards and follow up-to-date guidelines.

brennen• 13 years ago

It might be better than Perl, but not by much.

Oh come on.

PHP is more evolutionarily fit than Perl in a handful of very important ways (I would illuminate, for example, function signatures and a pre-built object system, syntactically trivial nesting of data structures, and a newcomer-friendly out-of-the-box execution model).

That said, I wrote Perl for years as a hobbyist (still do, now and then), and have spent the last 6 or so primarily writing PHP. There is not a day that passes that I don't miss some aspect of the rich, expressive, elegant, sophisticated, and intellectually delightful environment that is Perl at its best.

It is a doomed and sinking continent, Perl, but if we let the considerable wisdom of its once-great civilization evaporate into trite dismissal, we as a culture stand to lose a great deal.

kbenson3• 13 years ago

syntactically trivial nesting of data structures

I always found PHP's insistence on the array() keyword annoying, and the mixing of associative and sequential arrays ambiguous.

Perl's data structures (in their simplest form) and JSON are very similar, and I think JSON is a fine format.

You're mostly right on the other points though. Although I do love Perl's use of signatures in some cases.

girvo3• 13 years ago

I've always wondered if there was a place for a language that was properly designed, but copied PHP/php-fpm's execution model? It has beautiful horizontal scaling, but perhaps the long lived process model is better (ruby, go, etc) than PHP's "stateless" (as in, the app holds no real state itself after the request, it shuts down) per-req setup. Thoughts?

igul222• 13 years ago

See CGI.

girvo3• 13 years ago

Actually, I've been looking at SCGI with Nimrod... ;)

Seriously though, I've been trying to find out lately whether CGI is at all useful in this day and age, or whether if some well-designed scripting language that used the PHP style of run-time would be better to avoid using it.

There's not much info on CGI/SCGI in 2013 :'(

djmdjm• 13 years ago

It's amazing that, after all these years, PHP still has the ability to horrify me.

Hello again, old friend.

jamesbrennan• 13 years ago

Is the PHP source from 1994 available anywhere? That would help determine if this is Rasmus's satire or not.

mappu• 13 years ago

It's http://museum.php.net/

EDIT: It's true, it's not satire... check out e.g. lex.c in php2/php-1.99s.tar.gz

    /* Command hash table
     * The hash is extremely simplistic and just based on the length of the
     * command.
     */

followed by all top-level functions ordered by string length.

abimaelmartell• 13 years ago

http://museum.php.net/php1/

snyff• 13 years ago

from php-2.0.1/src/lex.c

void IntFunc(char *fnc_name) { int i=0; int cmdlen = strlen(fnc_name);

  while(cmd_table[cmdlen][i].cmd) {
    if(!strncasecmp(fnc_name,cmd_table[cmdlen][i].cmd,cmdlen)) {
      cmd_table[cmdlen][i].fnc();
      break;

fprawn• 13 years ago

There's a lesson here about how sometimes it's okay to fit your data to your code. This was, afterall, in php for years without causing much, if any, problems for end users.

I guess there won't be much agreement here on HN, but this deserves to be recognized as an amusing and clever solution to a problem. Even if the problem is self created.

PHP is great for doing quick and dirty dynamic web pages. So what if it doesn't scale out to a million line program?

awestroke2• 13 years ago

Except this was no acceptable solution. Instead of using a better hashing algo than strlen(!), the function names was chosen by length, leading to one of the largest and most criticised flaws of PHP: the inconsistent function naming.

RobAley• 13 years ago

PHP is great for doing quick and dirty dynamic web pages.

And a lot more besides (subtle plug for http://www.phpbeyondtheweb.com )

So what if it doesn't scale out to a million line program

I've seen more that one PHP codebase of that size, happily (and maintainably) plodding along.

dsego2• 13 years ago

Fu*k it, ship it?

ibudiallo• 13 years ago

Ok i will admit it, i didn't understand a thing. Would someone care to explain it to me?

dsego2• 13 years ago

I'm not really sure if this is right, but my best guess is that script parsers use a hash table to map the name of a function to a function pointer or something. Hash tables enable fast information retrieval by some key (usually a string). By just knowing the key you immediately (or after some calculations) know in which bucket to search for the element. For details see http://en.wikipedia.org/wiki/Hash_table.

For the hash table to be efficient, an algorithm that stores the elements has to evenly distribute them across the buckets. But instead of using a "proper" algorithm, a string length was used to determine in which bucket a function name belongs. To have the names evenly distributed, the names had to be of different lengths, even if that meant abandoning some naming conventions.

PHP functions originally bucketed by strlen, were renamed to balance length

Related Stories