Subject: Re: patch: SmartQuotesEnable preference item
From: WJCarpenter (bill-abisource@carpenter.ORG)
Date: Tue Jul 18 2000 - 18:21:45 CDT
tf> In short it isn't just about re-maping smart quotes, but also
tf> potentially about doing on the fly spell _correction_ and other
tf> actions that require us to to work as the user types. I think
tf> that putting this code into remapGlyphs is the wrong place to put
tf> it. Instead we should look to see where the squiggle code is and
tf> if you really want to, generalize that so that it can do
tf> squiggles, smart quote substitution, auto typo correction (ie teh
tf> --> the automatically), automatic outline collapsing and the like.
tf> That would seem to me to be a much better way to spend your
tf> development time.
Despite my earlier sketch of an example implementation based on
remapGlyphs(), I actually am in complete agreement with what you said
right there. I even started down a path of implementing it last
weekend.
Is the person who implemented the background spell checking still
around? Well, in any case, I've been through that code recently, and
I'd have to say it is a darned decent implementation.
A heartbeat timer fires every 100 ms, at which point a FIFO queue of
blocks-in-need-of-spell-checking is looked at. The block at the head
of the queue gets spell-checked (with all that attendant squiggle
stuff) and then removed from the queue. Various other parts of Abi
just cause newly added blocks (from typing, from pasting, etc) to be
added to the queue. For me, the user experience is quite nice. I
don't know how 100 ms was computed, but it seems to work out right.
When you type a misspelled word and move on to the next word, you have
the illusion that Abi is doing "check the word I just typed", but it
is really doing "queue for spell checking the block I just finished".
Last weekend, I began the work of generalizing this mechanism (with
the view of using it for smart quotes somehow). To maximize current
confusion in the interests of minimizing long-term confusion (among
developers), there is a massive, selective variable and function
renaming to reflect the concept of "background checks" rather than
just "spell checks". Spell checking then just becomes one instance of
a type of background check for a block. The idea is that a block can
be queued for one or more reasons. When it gets to the head of the
queue, the list of "reasons why it was queued" is examined and a
handler is called for each.
Because of the tedium of renaming everything and sorting out the
spelling-specific stuff from the general mechanism, and because of
[lame excuses], I haven't yet finished this reworking. I intend to
finish it within the next few days and submit a patch for it before
later layering any smart quote stuff on top of it. (It's really easy
to test whether I've broken background spell checking! :-)
More good news: it's all XP code that changes and is actually limited
to a small handful of files.
-- bill@carpenter.ORG (WJCarpenter) PGP 0x91865119 38 95 1B 69 C9 C6 3D 25 73 46 32 04 69 D6 ED F3
This archive was generated by hypermail 2b25 : Tue Jul 18 2000 - 18:24:15 CDT