OpenFormula teleconference meeting notes 2009-08-18

["David A. Wheeler" <dwheeler@dwheeler.com>]

We had our formula subcommittee teleconference today, 2009-08-18.  Here are my notes; please "reply all" with any corrections, etc.  As a subcommittee, I don't think we're obligated to have formally approved minutes, but I think it's helpful to try to write down what happened in case we need that information later.   So here's what *I* think happened, to the best I can put it down briefly.

Attendees:
David A. Wheeler
Eric Patterson
Andreas Guelzow
Rob Weir
Eike Rathke
Dennis Hamilton
Doug Mahugh

We discussed the comment-resolution process, in particular when using JIRA.  Rob Weir will give everyone rights to modify status of comments, so that we can mark comments as "resolved" once the assignee determines a proposed resolution.  We need to make sure that people have a chance to object to any proposed resolution, but we also want to make sure we don't waste time - we should only spend time on stuff where there are issues.

Next, we tried to go through the comments resolved within the last few weeks.  The big "lesson learned" is that Wheeler needs to define a publicly-accessible JIRA filter, before each teleconference, so that everyone can see the same list.  In the last two weeks, several folks have painstakingly walked through the document to verify that they are "already fixed" and closed them out (my thanks to those who tracked them down to make sure they'd been resolved!).  If a comment is to be deferred to a future version of the specification, do NOT mark them as "deferred"; instead, assign them to ODF-Next, and remove any assignment of a person.

There was a longer discussion of OFFICE-1187, which focused on "convert to NumberSequence".  The problem seems to be that when someone looks up "NumberSequence" as a type/pseudotype, they don't find it. NumberSequence is intentionally not listed as a type or pseudotype because there is no requirement that spreadsheets be able to store sequences inside a cell. Sequences are normally implemented by iterating through the cells of their ranges and determining which ones meet their criteria (e.g., which ones contain numbers).  The group proposal is to add headings in types or pseudotypes for sequences, which then cross-reference to the "Convert to..." text. That way, people can find the information when they look up types.  Wheeler will write the proposed text to connect these together. (SC-RESOLUTION-ACCEPTED 2009-08-18).

The group then discussed who will accept which unassigned comments (we've worked through a large number in the past two weeks, time for more assignments).  Each member is responsible for marking the "assigned to:" on a comment when they accept an assignment.  Here are some of the assignments accepted:
* Rob Weir will pick up all the "Re: " comments, as well as several financial functions such as IPMT, 1121, 1113, RRI (1106 and 1107)
* Eric P.: RATE, 1102, YIELD (1103 and 1104), 1895
* Doug M.: 1095, 1099, 1101
* David: 1120 (he has several others, inc. the whole basis issue)
* Eike: 1989
There are probably more assignments, but since members who accept assignments will enter that information into JIRA, I didn't feel the need to track of that - the purpose of JIRA is to track that sort of information for us!

We ended with a brief discussion on character encoding (UTF-8/16/32, JIS, combining characters), the "B" functions like RIGHTB, and so on.  There is a major question as to how we should deal with character encoding issues.  Eric P. is picking up 1895 and will come back by the next teleconference with at least more information on how Excel deals with these, and then discuss what should (and shouldn't) be specified.

--- David A. Wheeler