This change revisits commit eb7fa6f066. I think the problem
is not in the rendering of index entries but on the generation
side. The generator was converting Xexprs to strings, except
that strings were left as-is, which is inconsistent. So,
I've reverted commit eb7fa6f066 and fixed the generator.
Noted by ozzloy on IRC. The search page appears to fail when
searching for the term 'string<?'; in truth, the search succeeds, but
the search link-printing code doesn't escape the '<?' part of the
identifier.
* Stamp "plt-index.js" with the path of its generator.
* Fix most of the "use strict" and js2-mode warnings in scribble's
JavaScript.
* Some code improvements in the generating code too.
(With some edits by Eli.)
Revise the subword matching to have three different levels for full
match (= a permutation of the input), a generic match for all parts, and
everything else.
Instead of just one array for exact results and one for all the rest,
use one array for each possible comparison result, and concatenate them
all for the final list.
"quoted" things so "}" can be included in these. Not documented at the
user level for now, since I'm not sure that this will work fine in
general, but it's good enough to reactivate context-sensitive searches.
svn: r16944
anything for speed. So dump all attempts at making things faster this
way. Also, it seems that things are responsive enough with any break in
the searching "thread", so shorten that time. (And one other random
optimization: abort early when an item doesn't match.)
svn: r16943
* abstract html preference boilerplate code
* added help blurbs on prefs
* reorganize search code (better for adding tests in the future)
* added some "test cases", basically queries to try
* added pre-query that pre-filters the data
svn: r10127
