[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [oic] discussions on plugfest mailinglist
Hanssens Bart schrieb: > Hi, > > for those not on the The Hague plugfest mailing list, a short summary of what is currently > being discussed: > > - writing out "default" values like page margins: should implementations be required to write > them out (probably yes) and/or should the spec itself contain default values (probably not ?) I think it depends on the type of the value. If the "default" value is arbitrary, it should be written out. Page margins are such an example. There is no natural default for page margins, so write them out always. Applications may know and handle some default internally, but these defaults may differ in other apps. But there are other cases where a default value can be assumed easily, which should be specified then. Especially if that saves file size for mass objects such as spreadsheet cells. Writing out some defaults also helps some "simple" viewer tools that do not want to know all the defaults for various values. A similar case that comes into my mind are results of sheet formulas which are written out besides the formula. A viewer can use that result to render the sheet, without having to parse and interpret the formula. > - fixing implementation bugs (most notably the trailing slash issue when embeddeding > objects) : some implementations do it correct (as mentioned in the spec), others do not. > For the sake of interop, should conforming apps accept non-conforming documents when > it's trivial to implement this ? Balance between proliferating bugs and real world interop. They should of course. This helps to increase interoperability (which is one of the goals of this TC, isn't it?) :-) Developers make mistakes, that's a fact, and it will always be so. To improve the experience of the *users* of our ODF apps (well, we make the apps for the users, right?), the apps should do the best to understand the ODF written by other apps. I can even imagine to add comments to the spec describing real-life behaviour, it would help developers using the spec to implement interoperable *and* conforming applications. Daniel
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]