Crunch…

At home, I use the Amaya browser/editor to do the bulk composition of web pages (fiction and some pre-composing of blog entries before I dial up). One of its handy features is that it comes with a spell-checker (which has to be taught en-GB, alas). Unfortunately, I seem to be able to hit on words that cause it to crash (well, vanish like a popped soap-bubble).

Some while back I found that the misspelt "superintendant" (should be "-ent") crashed it, and that was fixed. Then in recent days I found that I'd managed to compose a document that had 4 separate space-delimited strings that caused version 8.4 to silently become one with the snows of yesteryear.

• superhurricanes
• supersymmetric
• virtualization
• underambitious

I suspect that there must be something anomalous about the handling of the "super-" prefix that the first time around didn't fully address.

Now version 8.5 is out, only a couple of weeks after 8.4, and is reported to address this issue and other crashes. I wonder how long it will be before I manage to find another word it chokes on. I could try supercalifragilisticexpialidocious, but that would be cheating - I'll just wait to find things that come up naturally as I write.

Later - I spellchecked this entry as a regression test. As expected, the regression strings passed OK, but Amaya 8.5 reports that it has no match for "supercalifragilisticexpialido" and separately for "cious". I spot a fixed length buffer (30 characters including a nul?) being used rather than the check being made on a range of references into the document.

Conversely I also spotted that if a word contains markup e.g. the source looks like

compl<em>i</em>ment

that it is taken as separate strings delimited by the markup as well as the surrounding whitespace, which is what I'd expect from a single reference range check rather than through any assembly of the word into a separate buffer.