module Jumplinks

[adrian]

I haven't had a chance to play with this yet, but I know it will be used on an upcoming site - just wanted to say how great it looks and most importantly how impressive the documentation looks to be - love the use of Github wiki!

[Mike Rockett]

You were probably going to have countless issues if you'd played with it before yesterday as I didn't test across multiple PHP/MySQL versions. (I'm lazy like that.)

But all seems to be good now - lemme know if you bump into any issues.

Was going to host the docs myself... but the tools are there, so use them - hehe!

[Mike Rockett]

The mapping collection API has been modified:

1. Method has changed from addCollection() to collection() due to:
2. the fact that it now picks up if a collection already exists so that it can add data to it.

Docs updated.

(Thanks Dave.)

[Mike Rockett]

Update: No, something's gone wrong there. Don't update just yet.

Fixed.

Edited January 31, 2015 by Mike Anthony

[Mike Rockett]

Just fixed up a little issue regarding the first TIMESTAMP column being auto-updated.

Basically swapped the created_at and updated_at columns around, and removed the updated_at = CURRENT_TIMESTAMP bit from the SQL template used for updating mapping collections.

I could have just defaulted to NULL for both of these timestamps, but I think this is better. Who cares about column order, anyway?

[Mike Rockett]

Okay, so after all those fixes, my testing seems to show that everything is stable. Everyone agree? I'd like to set this to stable in GH and in the directory.

[Dave Damage]

I will probably be able to test it on the weekend throughly, but so far it looks stable and works like a charm.

[Mike Rockett]

Are we ready to go ahead and make this stable? I haven't encountered any further issues...

[Dave Damage]

Are we ready to go ahead and make this stable? I haven't encountered any further issues...

Hell yeah. I didn't encounter any further problems, but I haven't done the big test yet. 

[Mike Rockett]

Cool beans - lemme know if anything comes up.

[LostKobrakai]

I've just installed the module, but it seems the redirect is to late and I get a "Cannot modify header information - headers already sent by" error.

Edit:

Changing the hook from addHook() to addHookBefore() solved the issue.

[Mike Rockett]

I've just installed the module, but it seems the redirect is to late and I get a "Cannot modify header information - headers already sent by" error.

Edit:

Changing the hook from addHook() to addHookBefore() solved the issue.

What sent headers first?

[LostKobrakai]

My templates _init.php, but even as I removed that part the rendering of mustache in a appended TemplateFile was earlier than Jumplinks.

[Mike Rockett]

I see. So there could be a lot of scenarios involved. Think it would be worth it to change the hook to something way early, and then check to see if a page representing the request exists? We could also add a configuration option that defaults the hook to pageNotFound, but allows you to specify if something is rendering output before the 404. Would be a simple checkbox that says, "Check Jumplinks in an early event - this prevents interruptions from other hooks/templates/etc."

What do you think?

Side note: my Redirects module for Bolt ties into a very early app event. Sure, the caveat is that pages that exist with the request uri are ignored, but we could add a simple check. Perhaps it would also perform redirects faster, as less things boot up?

[LostKobrakai]

I think the hook is the right one, just addHookBefore does force template rendering to wait for your hooked function to finish, while addHook doesn't seem to do this.

[Mike Rockett]

Okay, I think that's best then. Will change it now now

[Mike Rockett]

Done

[Mike Rockett]

I think we're ready for stable. I'm first going to switch over to prepared statements - silly me for not doing this first. Once done, I'll switch over to stable.

[Mike Rockett]

Version Bump to 1.1.0

Module is now stable (it looks to be, at least).

1. Now using PDO Prepared Statements
2. Tables honour responsiveness now - there was an issue with tfoot, so I just hid it from mobile...
3. Various bug fixes
4. No schema changes

Let me know if you bump into anything weird. 

Oh, and there's this issue - anyone have any input on that? (Personally, I prefer to have full control... Don't really like modules making decisions for me.)

[Mike Rockett]

Version Bump to 1.1.1

1. Using minified JS (manual call)
2. Refactor SQL object
3. Various other changes/fixes
4. No schema changes

(And GitHub is under attack - boy, how I battled to commit the changes... Finally slipped through after about 20 minutes of trying.)

[Mike Rockett]

Version Bump to 1.1.2

1. Correct page-check behaviour (was using name, now using URI)
2. Use commitJumplink in CSV importer
3. No schema changes

[Mike Rockett]

Version Bump to 1.1.3

1. Error-control for REQUEST_URI (previously threw an exception regarding it being undefined when using wireshell)
2. No schema changes

[ceberlin]

I actually had to uninstall the module - it was causing excessive mysql database connections (not usage) and my webserver nearly stalled.

Maybe I run into a loop or something?

I was redirecting from an old Drupal website to PW and used about 70 simple type redirects, such as sources  /drupal/page1/   

Got this (on the latest PW dev)

Error:

Exception: SQLSTATE[HY000]: General error: 1116 Too many tables; MySQL can only use 61 tables in a join (in /wire/core/PageFinder.php line 300)

Error:

Exception: DB connect error 1203 - User already has more than 'max_user_connections' active connections (in /wire/core/Database.php line 68)

I cannot say 100% that this was Jumplinks. I just noted that the performance was back to normal after I uninstalled the module.

So this is just for information in case anyone else can confirm this.

[Mike Rockett]

Thanks for your post.

Looks like another module is doing that. Jumplinks only creates two tables (one for jumplinks, and another for mapping collections).

In terms of performance, I'm unable to test that sort of thing. Are you saying that those errors go away when you uninstall Jumplinks? Do you have the same issue if you use ProcessRedirects?

If you (or anyone else) could provide me with more information, that would be great. Otherwise, I can't determine what's going on.

[Mike Rockett]

On the topic of performance, I'm now wondering if I should set up a cache system for Jumplinks. It was mentioned by Antti a while back that Jumplinks is already feature-rich, but I think that caching redirects will actually be a life-saver for some users with large sites. At the end of the day, running the regex engine for every incoming request is sure to be a big performance hit on any server (the way I see it).

I'm sure everyone agrees this should be added?

There are also a few other fixes and enhancements in the pipeline, the most important one involves optional query-strings (much like the QSA flag for mod_rewrite). At the moment, if you want to match something.php and something.php?foo is requested, the redirect will not occur. In order to get it to work, you'd need to add {all} on the end of the source URI, which is, quite frankly, not the best way of doing it.