Date: Mon, 28 Apr 1997 14:55:29 +0200 (MET DST) From: Arno Hollosi Subject: Re: RE and Date formats David writes: > You have made me think about the user interface some more, and > here is my plan for RE and DT formats now. > For RE, I will present a set of radio buttons, similar to YAGO > > and I will also show a read-only field with the text from the > file's RE property value. This field will be blank and greyed > > For date I will do something similar, giving a set of MM/DD/YYYY > data entry fields, and a text box for entering more information. This looks like a good compromise. > I'm imagining that someone adds an arrow to a game record, the later he > selects 'save' and a dialog box pops up "you must first correct the > date to ISO format" :) Ok :) Forcing the user to change date isn't really a good idea. A note about the required format for DT: - partial dates are possible (e.g. 1849, 1903-04, ....) (I'll add this to the spec) - games played on more than one day: - for two days use 'date1,date2' where year and/or month may be omitted in date2 (if they are the same as in date1) Examples: [1997-04-25,26], [1997-03-27,04-01], [1996-12-30,1997-01-02] (this is already in the spec) - for more than two days: just specify the first and the last day the game was played and use ':' as delimeter, e.g. [1997-02-10:03-12] Year/month may be omitted for end date if they equal the beginning date - maybe we could even specify a sohisticated format like the one Robert mentioned, i.e. combination of ',' and ':' within the simpletext value. But this might be a little bit too much of a good thing. I think that the first date is the most important (beginning of game). If this is in ISO format I don't care about the rest, i.e. it might be bad style, but I would've no problem with: DT[1997-04-26 16pm-22pm, rainy afternoon] Because tools need only one date for sorting/searching etc. And this should be the date the game started. Robert Jasiek writes: > Where to save non-ISO date texts (like the above or like an Edo > date) so that the program knows that it is to be shown in an > accompanying date text field? I can think of three solutions: > - format expanded to: (list of [date] | [Adate:Bdate]) | simpletext This is a nice solution but incompatible to FF[123]. I don't think that it's worth going through this. > - format only: simpletext. Try to parse as > (list of [date] | [Adate:Bdate]) else interpret as simpletext. As mentioned above: maybe it's best to state that DT[] may be followed by any text after an ISO date. > - second date property ;DC of value simpletext for date comment No old application could parse this property. That's a disadvantage. It's a good idea though. I'm still not sure what to do about illegal date/result fields. Anders & David suggest to leave them where they are. (maybe the application should at least notify the user that these values are incorrect [only if he/she edits those fields - otherwise such a message would be too annoying]). Other opnions? /Arno