Re: [PHP4BETA] Installation quirk
| From: | Zeev Suraski | Date: | Tue, 01 Jun 1999 22:00:21 +0000 |
| Subject: | Re: [PHP4BETA] Installation quirk | ||
| References: | 1 2 3 4 5 | Groups: | php.version4 |
| Request: | Send a blank email to php-version4+get-1240@lists.php.net to get a copy of this message | ||
At 22:26 01/06/99 , Sascha Schumann wrote:
On Tue, Jun 01, 1999 at 10:47:01PM +0200, Zeev Suraski wrote: At 14:41 01/06/99 , Sascha Schumann wrote:You said you don't see your ISP doing this upgrade. I pointed out that this sort of procedure is required for just about any other component in the server platform, so unless your ISP sucks, he does that sort of stuff anyway.To maintain backwards compatibility no version of PHP should claimx-httpd-phpfor itself. Using the version numbers in the MIME type to distinct between older and newer versions of PHP should make it easier for IPPs to provide even PHP version x and x+1. I don't think any provider will jump right on using PHP4. Experiencetells youto wait before using any new major release in a productionenvironment. Thisis true for all products, PHP4 will be no exception there.You can rarely run products side-by-side and migrate. You can't do it with Apache. You can't do it with ftpd. You can't do it with mod_perl. You can't do it with just about any server software. Because the mass of software does not allow parallel execution, we cannot do it either? That is like saying, because the major number of computers tend to crash daily, there must not be a single computer which has an uptime of more than two days
In a previous post you pointed out that "Hundreds of thousands of people still use Apache 1.2.x because upgrading to 1.3.x gives them no noticeable gain.I know I did. Apache 1.3.x doesn't indeed give the average user any noticeable gain over Apache 1.2.x. That's not the case with PHP 4.0 vs. PHP 3.0.
I swear that you will convince these people to upgrade by providing what Apache is lacking so apparently.Didn't quite follow you here...
My point is that you will get more people testing PHP4, if you provide this parallal functionality. It is not a must, it is an additional feature which will make every admin happy. "Oh see, we can just put it on our machine and it will not conflict with our other software. We even can switch site by site over to PHP4.My point is that there isn't a single program I've come across that has its major version number tagged on its functions or constants. A million coders are bound to be right. See below.
The upgrade process in these environments is usually running the server on another port, or on another machine that has access to the filesystem(s) in question, configuring it as much as you can, testing it through a short 'beta', and then going live. PHP should be no different. You always have the CGI alternative, too. CGI is dog slow.That's a myth. It simply is. Facts are that unless you have a truly busy site (i.e., hundreds of thousands of hits a day) - CGI is a walk in the park. When you consider PHP 3.0, the startup overhead of a fork() and an immediate exec() of an image that's already cached in the disk cache is minimal, and is pretty much insignificant in comparison to the runtime overhead of PHP 3.0. That's why with anything beyond simple scripts, PHP 3.0 simply isn't efficient - the startup gain cancels out VERY quickly. When your program is extremely quick, and only then, does the CGI overhead play a significant part in the overall overhead of the system, and even then, it's only when you consider percentages.
So, while bogging down the web server with huge numbers of CGI instances, we either have to accept that a) PHP3 becomes even more slower or b) that the speed gain of Zend/PHP4 will simply vanish. Both scenarios are indiscutable for sites getting more than a few hits per minuteAs I said, (a) is pretty much insignificant except for a very very busy site, and (b) is simply baseless. The speed gain of PHP 4.0 doesn't come from a quicker startup, but from a blazing fast execution engine. In CS, CGI overhead is O(1), or in other words, pretty much insignificant. Zend's improvements are somewhere between O(n) and O(n^2). In simple scripts, people won't see much of a difference (slightly slower performance due to CGI use, but otherwise, PHP 4.0 is roughly as quick as PHP 3.0) and with complex scripts they'd see a huge performance increase.
IPPs not wishing to be on the "bleeding-edge," and not wanting to risk to lose their customer base will wait until PHP 4 has been proven to be stable. For these sites it would be a big win, if it would be possible to provide PHP4 additionally without playing tricks (such as redirecting, moving sites to other hosts etc.)They'd easily use the CGI. There are simply no reasons against it.
Again, it is an additional feature to maximise the size of the user base of PHP4 as soon as possible. More users, more testers, more bug reports, better code. And think of zend.comI do think of PHP 4.0's success as well as on zend.com. What I don't want to give people is the impression that PHP 3.0 -> PHP 4.0 requires a migration process, because it really doesn't. It's so compatible we didn't even bother to make them both run side by side (let alone the technical difficulties and ugliness of code we'd have to patch if we were to make them run side by side). And this isn't marketing crap, this is really how it is. Zeev -- Zeev Suraski <zeev@zend.com> http://www.zend.com/ For a PGP public key, finger bourbon@netvision.net.il