New Forum Issue: Links via Google/Other Outside Resources

DavidZemon

Old links to the forum no long function correctly, and only direct to the new forum home page. Google will eventually get this fixed over time, but hand-written links from blog posts, emails, other forums, facebook, etc are all toast.
For instance, the link to my beloved PropWare thread is now dead :'(http://forums.parallax.com/showthread.php/157005-FYI-PropWare-Complete-build-system-and-library-for-PropGCC

Bump

Odd, that is actually an issue with the redirect; I'll look into it.All showthread.php/######-Stuff-Words-Things should go discussion/######/Stuff-Words-Things
The link you posted above seems like it is about to, but then stops and redirects to the homepage. I'll figure out why.

jmg

Odd, that is actually an issue with the redirect; I'll look into it.All showthread.php/######-Stuff-Words-Things should go discussion/######/Stuff-Words-Things
The link you posted above seems like it is about to, but then stops and redirects to the homepage. I'll figure out why.


I think the numbers have changed.
Google suggested the original  showthread.php/123170 was what i wanted,

but quite some digging finds this
http://forums.parallax.com/discussion/comment/992526/#Comment_992526

with this as a root

http://forums.parallax.com/discussion/123170/propbasic-reciprocal-frequency-counter-0-5hz-to-40mhz-40mhz-now

and this base-link seems to work
http://forums.parallax.com/discussion/123170

so maybe the trailing punctuation is an issue ?

Phil Pilgrim (PhiPi)

I rarely post links with verbiage beyond the first ######, since it's not required. If pattern matching requires the extra stuff, it will fail with my links.
'Just now noticed my avatar: apoplectic purple with an expression of incredulous astonishment. I think I'll keep it until things get ironed out -- or better yet, rolled back. 

-Phil

GordonMcComb

Well now, we're finally getting somehere! I'm relieved. This should be easy to fix, both in Apache and in the HTML parser.
The original thread has this URL:
http://forums.parallax.com/showthread.php/123170-PropBasic-Reciprocal-Frequency-Counter-(0.5Hz-to-40MHz)-40MHz-now-!!!

Whereas the Vanilla version maps to this:
http://forums.parallax.com/discussion/123170/123170-PropBasic-Reciprocal-Frequency-Counter-(0.5Hz-to-40MHz)-40MHz-now-!!!
or 
http://forums.parallax.com/discussion/123170/propbasic-reciprocal-frequency-counter-0-5hz-to-40mhz-40mhz-now

The first variation might be easier to do in RegEx because it merely replaces showthread.php with 'discussion,' picks up the thread number which is at the start of the page URL. Vanilla specifies the canonical name of the thread in source (the latter example I show), so there's no duplicate content penalty. This is actually a VERY nice feature.

GordonMcComb

Oops! Looks like the forum only accepts the canonical version as clickable links, I'm guessing because of disallowed characters in the link. Either the software will need to allow these characters, or any on-the-fly HTML parsing will have to first fetch the canonical name of the page in order to get the sanitized version, as it may not always be easily deduced. That's a hit to the DB that will slow things down.
As a point of reference, the links above are not stored as [link]http: / /someurl...[/link], but are parsed as links every time the page is generated. So changes to the acceptable characters in a link need to be done at the parser level. I don't see how these characters (e.g. parentheses) can pose a security issue, as they're pretty common in URLs. Vanilla just doesn't like 'em.
An alternative might be to encode the ( and ) characters.

So close but yet...

jmg

Because the original-base-number derived link seems to work fine why not simply truncate the rest, and use that base number ?

GordonMcComb

Just testing:
http://forums.parallax.com/discussion/123170/123170-PropBasic-Reciprocal-Frequency-Counter-(0.5Hz-to-40MHz)-40MHz-now-!!!


Okay, if parsing embedded links in the old posts -- which I highly recommend for reasons I've already mentioned -- some characters will need to be encoded, namely ( and ), and perhaps others. I'd do a test to see which characters Vanilla likes and dislikes before writing any RegEx code for this.