magnus |
08-22-2007 08:02 PM |
Quote:
Originally Posted by Jafo232
One being, there is only one plugin, and it uses one hook: url_to_bbcode.
|
One plugin location vs two.. heh.
Quote:
Originally Posted by Jafo232
Second, it uses Vbulletins built in method for matching URL's, instead of the one built into this hack.
|
Correct, because if you use any other hacks that toy with vB's internal URL parsing routines it will most likely break this hack.
Quote:
Originally Posted by Jafo232
If the user chooses not to parse links, during a reply or threadpost, then the hack won't override it as the hack you linked to appears to do.
|
This is true, however, because it's by design. From an Admin's perspective, this mod is a great SEO tool. It's mostly for the admin, not necessarily the users.
Quote:
Originally Posted by Jafo232
I have never heard of a host disabling fopen, the only thing is when it used to open FILES, and safe mode is on, it will check the UID. However, I do not believe this applies when calling a remote document via http.
|
Well, just to name one.. DreamHost. fopen() is generally considered a security risk on shared hosting servers.
Quote:
Originally Posted by Jafo232
URL on the other hand, must be installed on the system (well at least the libcurl package) whereas fopen is native.
|
You'd be hard pressed to find a shared hosting provider that doesn't have cURL available. And those not using shared hosting can simply install it themselves. But I digress..
Quote:
Originally Posted by Jafo232
What happens with the other hack when a user say, posts a link to the latest fedora distribution that is 2.7 GB? From what I see of the other hack, the server will attempt to download this entire package before parsing out a title (that it will never find). Of course, it will time out, or give you an exhausted memory error. My mod by contrast, only downloads the first 1000 lines of any document.
|
Well, the first problem here is that someone is trying to download Fedora.. when obviously they should be downloading a superior distribution such as Debian ;). That being said, there's a nifty (.. and configurable!) timeout option. By default, if a page takes longer than 5 seconds to return a title then the process is aborted and the default URL is used.
Quote:
Originally Posted by Jafo232
This also works with shoutbox or any other plugin that does not bypass the normal posting process, whereas the other only works with the newpost and and edit post hooks.
|
So.. I can hijack a shoutbox and cripple a server by spamming a bunch of links to 10GB ISO's and waiting for it to parse 1000 lines each time I paste it? However way this hack is implemented, by nature it's going to create additional load on the server. Why would you want to add an additional load to something that already puts a ridiculous strain on a server (re: Shoutbox)?
If you're going to provide a link in the OP to compare the differences, the least you could do is make sure the statements are factual.
|