Richmond goes data-mining (a.k.a. shovelling through the sh..)

[J. Landman Gay]

Richmond Mathewson wrote:
>  On 27/04/2010 18:17, J. Landman Gay wrote:
>> Peter Alcibiades wrote:
>>
>>> So its worth doing, and Richmond done good to make a start. 
>>
>> I made a list last night and will post it later today. There are some 
>> errors that the list can correct, but for the most part I think it is 
>> accurate.

I've now double-checked some of the things I wasn't sure of, and I think 
my list is pretty accurate now.


> Can you tell me how you did it? I have a feeling the way I went about it 
> was
> extremely and unnecessarily long-winded.

I extracted the dictionary entries by reading the custom properties from 
the doc clumps (the dictionary builder that ships with the MC IDE does 
this, so I just modified that script a bit.) Once I had a complete list 
of all 1600+ entries, I used BBEdit and grep to remove or extract 
various combinations of platform support.

I first removed from the list all entries that were available on all 
platforms, leaving only those that had at least one platform missing. I 
also removed entries where the dictionary is wrong (for example, all 
inks are fully cross-platform now but the dictionary hasn't been 
updated.) I also removed all synonyms (mostly the XBrowser references) 
and all instances of "COM:" except for the first one.

What I found when I was done was that the only tokens that Linux does 
not support fall into three basic categories: browser xcmd, 
quicktime-related, and things the OS itself does not support. I also 
found 32 Linux-only commands unique to that OS which no other platforms 
support. There are Linux/OS X commands that Windows does not support. 
There are 16 Windows-only tokens that don't apply to any other OS. And 
so forth.

The results are here: <http://jacque.on-rev.com/codebits/tokens.txt>. It 
looks to me like linux support is well-balanced, and that every OS has a 
good number of platform-specific tokens that can't possibly be applied 
elsewhere.

-- 
Jacqueline Landman Gay         |     jacque at hyperactivesw.com
HyperActive Software           |     http://www.hyperactivesw.com