[Home]AnnotationWikiIssues

MeatballWiki | RecentChanges | Random Page | Indices | Categories

CritLink is so cool, why hasn't it taken off? I don't think people don't want annotations. possible reasons:

So, I guess I'm conducting a poll. What do you think stops you from using CritLink? Do people just not want annotations? Why isn't the CritLink mediator used more widely?


Technical issues below

The bogus annotation problem

Since the only way to get an annotation out of crit.org's database is to change the [AnnotationWiki:AnnotatingPage], obviously many bogus annotations threaten to pile up. Details:

One thing we may eventually need if the idea catches on is another host of a CritLink proxy. The problem is that, by using crit.org's proxy, we have no way to remove non-wiki-controlled annotations (i.e. pretty much all of them) from their database. Worse, there are some small issues with the current CritLink software that need to be remedied. An example: on the page http://crit.org/http://bayle.stanford.edu/wiki/wiki.pl?AnnotationWiki, there are THREE annotations to the word this. Here are where they point to:

http://www.seedwiki.com/page.cfm?doc=AnnotationWiki&wikiid=368

http://www.seedwiki.com/page.cfm?doc=AnnotationWiki&wikiblurbid=25155&wikiid=368

http://www.seedwiki.com/page.cfm?doc=AnnotationWiki&wikiblurbid=25155&wikiid=368&edit=yes

As you can see, two of these annotations were entered into crit.org's database when i viewed some previous versions of SeedWiki:AnnotationWiki. Now that they are in that database, there are only two ways that i can think of to get them out:

  1. put a page at "http://www.seedwiki.com/page.cfm?doc=AnnotationWiki&wikiblurbid=25155&wikiid=368"; that does not contain a <link> tag, and view it with crit.org's Mediator
  2. have an administrator at crit.org remove the link from the database there.

#1 may be practical in this case, since you run SeedWiki, but there will doubtless be many sources of bogus annotations on the web that we cannot control (i.e. google caches, wayback machines, other wikis, etc). A better solution is to run our own Crit Link Mediator server, and periodically run a perl script that erases annotations of a certain form from its database (e.g. any annotation that points to an URL containing "wikiblurb").

Wanted: a site to host a changing CritLink.

Or maybe KaPingYee, the owner of the crit.org site, would agree to filter some entries from his crit.org Mediator's database. No one's asked him yet. :) -- BayleShanks

You can't edit any AnnotatingPage? when viewing through CritLink

When editing an AnnotatingPage? (the controlling wiki page), edit it directly, NOT through CritLink. CritLink will preprocess the link tags in the edit box; even if you hit save without making changes, you will have garbled the link tags. So before you edit any AnnotatingPage?, make sure that http://crit.org is NOT the first part of the URL that you are looking at.

I think the solution to this problem may be to patch CritLink to not process stuff that is the initial text in an edit box.

The problem with my CritLink

The CritLink mediator running on my computer, http://bayle.stanford.edu/crit/nph-med.cgi/, has a big problem; when you try to see anything in the www.usemod.com domain, you get a message that you aren't using HTTP/1.1. Any ideas? I am currently running version 0.9.2 of CritLink mediator in CGI mode. Maybe CGI mode doesn't do HTTP/1.1 yet and i need to run in standalone mode? Or do i have Apache misconfigured?

Note that crit.org's CritLink mediator CAN view all pages in www.usemod.com.

Until we fix this, it won't be too fun doing annotations using my CritLink. I mean, we can't even annotate MeatBall!

(I stress that you CAN already annotate MeatBall using crit.org; its just slow and has the BogusAnnotations problem.)

Need to include bang inside tags on SeedWiki

Details later. See AnnotationWiki:AnnotatingPageExample.

A patch for CritLink that allows full CGI URLs

(By CGI URLs I mean URLs with stuff like"action=edit&id=blah" or somesuch.)

In version 0.9.2 of CritLink mediator, you are given the option to enter comments in enhanced text mode; the program detects and hyperlinks any URLs that you enter. However, it thinks the characters = and & are illegal in URLs, so it will truncate any URL containing these. See [CritOrg:CritOrgWiki] for an example. Of course you can ammend this manually by using (partial) HTML-mode like tested in [a manual work around]

Here's a [patch] to CritLink (crit-0.9.2) that seems to solve the problem. To apply the patch, change to the directory containing the crit-0.9.2 directory, put the patch file (cgi_query_links_in_etext.diff) in that directory, then type patch -p 0 < cgi_query_links_in_etext.diff.


Note that I am not a hardcore programmer and so I apologize if my patch does something stupid! I sent a message to crit-tech@crit.org to ask for their advice, so if there's a glaring error they'll probably let us know soon.

Also, this is my first time using diff to create a patch, did I do it correctly? I did this:

mv crit-0.9.2 crit-0.9.2-p
tar -xovzf crit-0.9.2.tar.gz
diff -C 2 crit-0.9.2/web/nph-anno.cgi crit-0.9.2-p/web/nph-anno.cgi  > cgi_query_links_in_etext.diff

-- BayleShanks

Note: This page is bookmarked and has annotations on [diigo]. -- FridemarPache


CategoryWebAnnotation


Discussion

MeatballWiki | RecentChanges | Random Page | Indices | Categories
Edit text of this page | View other revisions | Search MetaWiki
Search: