[pmwiki-users] RFC: Indexing categories and links

[Peter Bowers]

On Thu, Mar 4, 2010 at 7:40 AM, Petko Yotov <5ko at 5ko.fr> wrote:
> On Thursday 04 March 2010 06:20:04, DaveG wrote :
>> On 3/3/2010 2:58 PM, Petko Yotov wrote:
> I now see a problem of which I haven't thought before -- with other parameters
> like name=!Name or group=!Name the "!" means "NOT Name" so link=!Name may be
> confusing and a bad idea. :-(
> Especially if at some point we change the link= parameter to accept multiple
> page names like link=Page1,Page2,-Group3.* . Hmm...

I believe the vast majority of authors use the "-" for negation,
although I could be wrong.  But in any event, we end up potentially
breaking existing sites which goes against philosophy.

The BIG advantage of using "!" is that the same code can then be used
generically with all other prefixes.  You get additional functionality
and consistency and reliability/stability because one codeset operates
all prefixes.  So maybe the solution is to look for a markup that uses
the prefix but avoids putting the prefix at the very beginning of the
linkname...

What if we went with something like this:
   link=^!^mycat
-or-
   link=(!)mycat
-or-
   link=\!mycat       // kind of like escaping the !...?  [1]

Unfortunately none of those is particularly intuitive.  Using the
backslash to escape the ! is perhaps the best and it wouldn't break
anything existing because the prefix character in pagelists links is a
totally new thing...  Presumably for the ~ prefix or etc you wouldn't
need the escaping backslash -- but if someone wanted to define a
hyphen prefix then it would need to be escaped.  I think the idea has
at least some merit...

Other markup suggestions that are more intuitive and don't suffer the
same drawbacks?

-Peter

[1] ParseArgs() could be coded to handle this across the board without
breaking things -- just strip leading backslashes.  It could really be
a nice feature making ParseArgs() much more flexible...