SEF/SEO URLs

[igorw]

I would also say, "it's 2011" let's get over this and realize that "pretty urls" is a thing of the past.

I humbly disagree.

While URLs are becoming less significant for users, the web is moving towards embracing HTTP. This can be observed by the whole 'REST' movement, Symfony2 building strong on top of HTTP with caching and ESI.

So what does it mean? It means using the capabilities of HTTP. Such as alternate request methods (GET/POST/PUT/DELETE). Alternate request and response formats.

So why does this matter? The formats are specifically where this gets really interesting. Let's say you request a topic, but ask for a `application/json` content type (by specifying an accept header). The server notices this and returns the topic encoded as JSON. This could also be XML. Now, you can do the same for creating a reply to a topic. Make a POST request to (for example) `/posts/new`, with a request body of {"topic_id": "4", "body": "My new post..."}

Well, this adds an API to any board, allowing services to interact with it. And that, imo, is huge.

[callumacrae]

I support the use of clean URLs and would quite like to see them include in phpBB. On the other hand, of someone asks me to install any of the current SEO modifications on their board, I will definately refuse, even if I'm being paid by the hour. The fact of the matter is that even if it is a great idea, it's extremely tricky to implement into an already written solution.

So why does this matter? The formats are specifically where this gets really interesting. Let's say you request a topic, but ask for a `application/json` content type (by specifying an accept header). The server notices this and returns the topic encoded as JSON. This could also be XML. Now, you can do the same for creating a reply to a topic. Make a POST request to (for example) `/posts/new`, with a request body of {"topic_id": "4", "body": "My new post..."}

Well, this adds an API to any board, allowing services to interact with it. And that, imo, is huge.

Nah, POST request to /api/:format/topic/:id would be easier

[Lumpy Burgertushie]

I won't pretend to understand what all of that means. however, what I do know is that it means nothing to the end user. the end user cares not what the url is as long as it gets them where they want to go when they click on it. my point about pretty urls being a thing of the past is that people do not type urls in very much anymore and that the requirement to type them in is going to be less and less. Therefore, the need for the pretty urls is going away. The only real need for them was for search engines several years ago and that is no longer the case


robert

[callumacrae]

I won't pretend to understand what all of that means. however, what I do know is that it means nothing to the end user. the end user cares not what the url is as long as it gets them where they want to go when they click on it. my point about pretty urls being a thing of the past is that people do not type urls in very much anymore and that the requirement to type them in is going to be less and less. Therefore, the need for the pretty urls is going away. The only real need for them was for search engines several years ago and that is no longer the case


robert

What happens if you want to see where you're going?

Code: Select all

[url=http://example.com/topic-cheap-viagra]View my tutorial here[/url]
[url=http://example.com/viewtopic.php?f=64&t=2100309]View my tutorial here[/url]

~Callum

[Erik Frèrejean]

I won't pretend to understand what all of that means. however, what I do know is that it means nothing to the end user. the end user cares not what the url is as long as it gets them where they want to go when they click on it. my point about pretty urls being a thing of the past is that people do not type urls in very much anymore and that the requirement to type them in is going to be less and less. Therefore, the need for the pretty urls is going away. The only real need for them was for search engines several years ago and that is no longer the case


robert

What happens if you want to see where you're going?

Code: Select all

[url=http://example.com/topic-cheap-viagra]View my tutorial here[/url]
[url=http://example.com/viewtopic.php?f=64&t=2100309]View my tutorial here[/url]

~Callum

That only works if the topic title actually covers the topic itself. And someone using that way to spam won't use that topic title . Also with these kind of URLs you'll need a topic ID nevertheless and the rest of the URL isn't used and thus can be spoofed very easily.

[dandv]

I won't pretend to understand what all of that means.

I have a friend who says he won't vote, because he isn't well-informed enough about politics to make a good decision.

Abstaining, when understanding is lacking, is a perfectly reasonable choice.

however, what I do know is that it means nothing to the end user.

Well, how do you know that? It would be useful to see some examples.

I gave some examples of when the users really care about URLs in my previous post.

[dandv]

What happens if you want to see where you're going?

Code: Select all

[url=http://example.com/topic-cheap-viagra]View my tutorial here[/url]
[url=http://example.com/viewtopic.php?f=64&t=2100309]View my tutorial here[/url]

~Callum

That only works if the topic title actually covers the topic itself. And someone using that way to spam won't use that topic title . Also with these kind of URLs you'll need a topic ID nevertheless and the rest of the URL isn't used and thus can be spoofed very easily.

This is a good objection to URLs like:

Code: Select all

http://example.com/thread-12345/topic-title-that-can-actually-be-anything

The cleanest solution is to use genuinely unique URLs:

Code: Select all

http://example.com/thread/topic-title-that-is-actually-a-unique-id

This approach is taken by the MyBB user-friendly URLs addon. To solve the problem of multiple topics with the same title, a "URL uniquifier" function is used:

The Google SEO URL uniquifier is applied to URLs that would otherwise
not be unique (and thus result in threads that are not accessible).
Collision testing (for example for two threads with the same title)
is done only once, therefore the uniquifier must result in a truly
unique URL that can not possibly collide with anything else.

A good uniquifier needs to fulfill these two criteria:

1) contain the items unique {id}
2) contain punctuation that cannot occur in non-uniquified URLs

Early versions of Google SEO used {url}-{id} as uniquifier and
therefore did not fulfill criteria 2). This could lead to collisions
in rare cases, for example:

ID: 1, Title: Unique, URL: Thread-Unique
ID: 2, Title: Unique 3, URL: Thread-Unique-3
ID: 3, Title: Unique, URL: Thread-Unique-3 (same as Thread 2)

Thread 3 collides with Thread 1 (both are called Unique), so the
uniquifier is applied. This results in Unique-3. However, there
already happens to be a thread called Unique-3. Doesn't work.

With the new uniquifier {url}{separator}{separator}{id},
the uniquified URL will be Thread-Unique--3. Because the id is
already unique, and other URLs can't contain -- (title punctuation
is reduced to one single separator, not two), that makes the URL
as a whole unique.

[Lumpy Burgertushie]

which strangely enough is exactly what you get with the non-friendly urls. each one is unique.
there is no "collision" and there is really no duplicate content problems either.


robert

[callumacrae]

which strangely enough is exactly what you get with the non-friendly urls. each one is unique.
there is no "collision" and there is really no duplicate content problems either.


robert

That isn't a valid argument at all.

[Erik Frèrejean]

What happens if you want to see where you're going?

Code: Select all

[url=http://example.com/topic-cheap-viagra]View my tutorial here[/url]
[url=http://example.com/viewtopic.php?f=64&t=2100309]View my tutorial here[/url]

~Callum

That only works if the topic title actually covers the topic itself. And someone using that way to spam won't use that topic title . Also with these kind of URLs you'll need a topic ID nevertheless and the rest of the URL isn't used and thus can be spoofed very easily.

This is a good objection to URLs like:

Code: Select all

http://example.com/thread-12345/topic-title-that-can-actually-be-anything

The cleanest solution is to use genuinely unique URLs:

Code: Select all

http://example.com/thread/topic-title-that-is-actually-a-unique-id

You still need to tuck in a direct identifier to the topic somewhere to get around clashes (no matter how/where you "hide" this). You can't rely on only the topic title or whatever slug you plan to use.

ID: 1, Title: Unique, URL: Thread-Unique
ID: 2, Title: Unique 3, URL: Thread-Unique-3
ID: 3, Title: Unique, URL: Thread-Unique-3 (same as Thread 2)

Thread 3 collides with Thread 1 (both are called Unique), so the
uniquifier is applied. This results in Unique-3. However, there
already happens to be a thread called Unique-3. Doesn't work.

With the new uniquifier {url}{separator}{separator}{id},
the uniquified URL will be Thread-Unique--3. Because the id is
already unique, and other URLs can't contain -- (title punctuation
is reduced to one single separator, not two), that makes the URL
as a whole unique.

What happens if I create a topic and give it the title Unique- 3?

[Lumpy Burgertushie]

which strangely enough is exactly what you get with the non-friendly urls. each one is unique.
there is no "collision" and there is really no duplicate content problems either.


robert

That isn't a valid argument at all.

it wasn't meant to be an arguement, it was meant to make the point that phpbb already has unique urls.
if you make them "pretty" and then have to make them "ugly" again to make them unique. What is the point?

You are just going around in circles.


robert

[callumacrae]

which strangely enough is exactly what you get with the non-friendly urls. each one is unique.
there is no "collision" and there is really no duplicate content problems either.


robert

That isn't a valid argument at all.

it wasn't meant to be an arguement, it was meant to make the point that phpbb already has unique urls.
if you make them "pretty" and then have to make them "ugly" again to make them unique. What is the point?

You are just going around in circles.


robert

Code: Select all

viewtopic.php?f=64&p=13015808
topic/seo-urls-7

It's not all or nothing, you can compromise.

[Lumpy Burgertushie]

which strangely enough is exactly what you get with the non-friendly urls. each one is unique.
there is no "collision" and there is really no duplicate content problems either.


robert

That isn't a valid argument at all.

it wasn't meant to be an arguement, it was meant to make the point that phpbb already has unique urls.
if you make them "pretty" and then have to make them "ugly" again to make them unique. What is the point?

You are just going around in circles.


robert

Code: Select all

viewtopic.php?f=64&p=13015808
topic/seo-urls-7

It's not all or nothing, you can compromise.

Is that supposed to be "pretty" or 'friendly"?

that just really makes my point.

from a SEO standpoint, none of this matters.

from a user standpoint, that is no easier to remember or to type in than what we currently have.
I contend that the standard "pretty" urls are not much easier to remember or to type in than what we have now.
but my main point is that most people never type in a url anymore. I can't remember the last time I did it and I work online all day every day. I copy and paste or click on links etc. Even when I am creating web pages I rarely have to actually type a complete url other than just a simple http: // whatever . com


robert

[dandv]

You still need to tuck in a direct identifier to the topic somewhere to get around clashes (no matter how/where you "hide" this). You can't rely on only the topic title or whatever slug you plan to use.

ID: 1, Title: Unique, URL: Thread-Unique
ID: 2, Title: Unique 3, URL: Thread-Unique-3
ID: 3, Title: Unique, URL: Thread-Unique-3 (same as Thread 2)

Thread 3 collides with Thread 1 (both are called Unique), so the
uniquifier is applied. This results in Unique-3. However, there
already happens to be a thread called Unique-3. Doesn't work.

With the new uniquifier {url}{separator}{separator}{id},
the uniquified URL will be Thread-Unique--3. Because the id is
already unique, and other URLs can't contain -- (title punctuation
is reduced to one single separator, not two), that makes the URL
as a whole unique.

What happens if I create a topic and give it the title Unique-3?

The uniquifer is {url}{separator}{separator}{id}. The double separator is key, because all URLs are normalized so that two consecutive separators cannot occur. (The normalization means removing punctuation and %-encoded characters, trimming the URL to a maximum length by replacing its middle characters with an ellipsis, and uniquifying the URL).

So Unique-3 will be uniquified to Unique-3--1, if a thread called Unique-3 already existed.

[dandv]

Code: Select all

viewtopic.php?f=64&p=13015808
topic/seo-urls-7

It's not all or nothing, you can compromise.

Is that supposed to be "pretty" or 'friendly"?

that just really makes my point.

from a SEO standpoint, none of this matters.

We don't care about SEO. These URLs are human-friendly.

from a user standpoint, that is no easier to remember or to type in than what we currently have.

How so? Again,
http://example.com/forum/viewtopic.php?f=64&t=2100309
vs.
http://example.com/forum/pets/nice-urls

but my main point is that most people never type in a url anymore. I can't remember the last time I did it and I work online all day every day. I copy and paste or click on links etc.

Well, it's great that you happen to work in an environment where you don't need to type in URLs, and it's also great that you don't care that a URL like http://example.com/forum/viewtopic.php?f=64&t=2100309 doesn't tell you anything (you probably also enjoy rickrolls).

But there are many other phpBB users who do type in URLs, and I've already listed many such examples. There are also many users who do get information from looking at a URL. How can I help you understand that? When Firefox 4 removed that status bar that displays the target URL of a link, hundreds of people were so enraged that a petition was created to bring that status bar back.

phpBB is not for you or for me; it's for a community of users, many of whom understand the value of human-friendly URLs. In general terms, what makes you use your personal preference and assumptions of the behaviors of others, as a reason against implementing a feature? Even if I personally think localization is a terrible idea, I'm not arguing against localizing phpBB, because I understand that other users need that.