Last Week in NVDA (3 Nov)

There wasn't much noteworthy activity to report about last week, aside from NVDA 0.6p2 passing 10000 downloads, about which I already reported. The one noteworthy change in NVDA was Mick's introduction of an NVDA specific dialog which allows the user to edit the content of a cell in Microsoft Excel. This dialog appears when you press f2 to edit a cell. This is necessary because the formula bar in Microsoft Excel is inaccessible to NVDA. It is only a temporary solution until NVDA has display hooks which can track the text and cursor in the formula bar. Nevertheless, it does allow the user to edit cells in an accessible manner, which will be a welcome relief for some users.

For BRLTTY users, Samuel Thibault is now building the BRLTTY installer I reported about last week when he releases new Windows builds of BRLTTY, and the installer script is now in the BRLTTY svn repository. Thus, I have removed the installer files from the location I linked last week. Please download installers from Samuel's BRLTTY Windows builds.

Mick has started an article about what we will cover in the upcoming hack fest, which is less than a week away now! As always, there will be a lot of brain melting discussion, though we hope to do a little more coding this time. :) We are keen to involve other developers in these discussions, so let us know if you are interested.

Mick and I debated how to improve braille review. Currently, braille either moves the caret or the review position, but it is always tethered to one or the other; it cannot move freely. This is potentially undesirable in some circumstances. There are two possible solutions:

• Braille always moves the review cursor. The caret can optionally be moved when braille moves, but regardless, the review position will always be moved. The review cursor itself would never be displayed. The advantage of this is that no separate braille cursor or range is required. The disadvantage is that review bookmarks or the like would be required, as one might wish to move quickly between the caret and an arbitrary review point.
• Braille can move beyond either cursor, but either cursor can be moved to a braille position, either on demand or automatically. This requires the concept of an active braille cursor; i.e. whether braille is manipulating the caret or the review cursor. This probably allows for the most flexibility, but possibly makes fora steeper learning curve, as there is an additional cursor/range to consider.

We still can't reach a consensus on which is the better method, but we have ideas on how both could be implemented. :) We also discussed ways to work around some bugs in liblouis which are causing some annoying exceptions when text contains certain unicode characters.

Aside from this, we've spent a considerable amount of time writing presentation proposals for CSUN 2009, which we submitted today. Also, we've been trying to track down some annoying crashes with Firefox and NVDA that have been reported lately, but haven't had much success yet.