PHP pages

[Ardesco]

Hmm, have you double checked the file permissions? you may be showing too much to the world

 

[Jog On]

Hmm, have you double checked the file permissions? you may be showing too much to the world

I can't set them to 777 but the script directory is 755

 

[NickFitz]

Personally I firmly advocate nowadays that the underlying technology has no business being present in a URI.

So, rather than http://example.com/catalogue/page1.html or http://example.com/catalogue.php?page=1 I'd be setting the server up to deal with http://example.com/catalogue/1 and so forth.

The .html extension seems harmless, whereas extensions like .php, .asp or .aspx become a nuisance when you move to a different platform and are left with a shedload of redirects that you have to maintain for ever. But even .html is inappropriate IMNSHO, because a URI is supposed to identify a resource (or the location of a resource, in the case of a URL), and by including .html you're restricting it to identifying a specific representation of that resource.

Therefore I would argue that, although you may support URIs that include content-type information such as .html, the canonical form of the URI should be in the form that contains no information about the content-type of the resource, and certainly no information about something so completely irrelevant to the identification of a resource as the technology used (at the moment) to serve that resource.

Note that, if you do support the version with the type-denoting-extension in addition to the canonical version, you should take the additional steps necessary to ensure that only the canonical form is indexed by search engines, or you'll risk being penalised for having duplicate content, and also risk being forced to maintain the non-canonical versions.

RESTful Web Services by Richardson and Ruby is an excellent book on the correct use of HTTP, including a detailed discussion of URL design.

 

[Jog On]

Personally I firmly advocate nowadays that the underlying technology has no business being present in a URI.

So, rather than http://example.com/catalogue/page1.html or http://example.com/catalogue.php?page=1 I'd be setting the server up to deal with http://example.com/catalogue/1 and so forth.

The .html extension seems harmless, whereas extensions like .php, .asp or .aspx become a nuisance when you move to a different platform and are left with a shedload of redirects that you have to maintain for ever. But even .html is inappropriate IMNSHO, because a URI is supposed to identify a resource (or the location of a resource, in the case of a URL), and by including .html you're restricting it to identifying a specific representation of that resource.

Therefore I would argue that, although you may support URIs that include content-type information such as .html, the canonical form of the URI should be in the form that contains no information about the content-type of the resource, and certainly no information about something so completely irrelevant to the identification of a resource as the technology used (at the moment) to serve that resource.

Note that, if you do support the version with the type-denoting-extension in addition to the canonical version, you should take the additional steps necessary to ensure that only the canonical form is indexed by search engines, or you'll risk being penalised for having duplicate content, and also risk being forced to maintain the non-canonical versions.

RESTful Web Services by Richardson and Ruby is an excellent book on the correct use of HTTP, including a detailed discussion of URL design.

Thankyou for that - and have a banana

It's all working now so everyone please have a banana on me:








It's the same HTML/PHP page being served but the one element I'm split testing, be it the headline, a paragraph, a graphic, call to action, testimonial, price etc...


... is swapped out - the PHP script just shows the element I choose to split test i different versions to every alternate visitor so I can measure if a change to a page has a positive or negative affect on conversions or not and if that change is statistically significant.

It's all about the money old boy

 

[Jog On]

This is a good explanation of why I have been faffing around with PHP:

http://www.marketingexperiments.com/improving-website-conversion/ab-split-testing.html

 

[NickFitz]

FWIW, my presentation at the first BarCampLondon was entitled The Correct Use of HTTP, which was intended as a pun on the title of Magazine's album The Correct Use of Soap - the idea being to argue against fundamentally misguided approaches such as SOAP, in favour of a RESTful approach that used HTTP as originally intended.

Sadly, nobody else there was either old enough, or sufficiently well-informed about HTTP, to get the joke