Peter Crowther <Peter.Crowther at IT-IQ.com> writes:
> I'd like to propose two ideas:
> 
> 1. Find some way of making the archive of fixes searchable by
> keyword.
>
> 2. Once this is done (or earlier, in the hope that it will be done!), ask
> developers of new fixes to put a line of likely keywords in the preamble.
> This may not even be necessary, as we may find that the existing keyword
> search works well enough --- an advantage of distributing source code!
> 
> Thoughts?

We already have a great number of keywords in the bodies of the mail
messages that announced these fixes. Before computers were ubiquitous,
scientific data was indexed in expensive abstract volumes that
assigned keywords to each publication. Now we have full text searches,
and I find them *much* more useful. I think the advantage is that
everyone used the same small set of keywords, making it very difficult
to be specific about what I wanted. Where authors did use specific
keywords, I was unlikely to guess what words they would use. Full text
searches are easier because I don't have to guess which specific word
the original author blessed, I just have to guess some words that they
would have had to use in their text.

Don't add keywords; search full text.  The more text we can make
searchable, the better. In particular, I think the mail message that
accompanied the submission, the preamble, the class comments, and the
source text are all useful grist for the search engine.

-- 
Stan