Two links people might find interesting in this context

Posted Dec 10, 2006 8:51 UTC (Sun) by khim (subscriber, #9252)
In reply to: Two links people might find interesting in this context by fergal
Parent article: Ecma International Approves Office Open XML as Worldwide Industry Standard

I hate to find myself defending MS but you can't really criticise MS for code reuse rather than writing it from scratch.

You don't write importers from scratch. You take library which deals with existing parts (CSS, SVG, MathML, XLink and so on) and write only relatively thin unshareable part. But since OpenXML does not use any existing parts (except the XML itself - and even then it does it wrong) you are forced to write the whole importer as one big blob...

Two links people might find interesting in this context

Posted Dec 10, 2006 11:06 UTC (Sun) by fergal (guest, #602) [Link] (2 responses)

I'm not defending OpenXML. You say the fact that they're reusing the code for Mac shows how complex OpenXML is. I say that reusing the Windows code for Mac tells us nothing about how complex OpenXML is. Except for the most trivial library, reusing code across platforms is simply the sensible thing to do. If they had implemented it from scratch for the Mac that would tell us that their Windows code was so bad that it couldn't be reused. I just don't see how not reimplementing from scratch for Mac can be used as a negative point.

Two links people might find interesting in this context

Posted Dec 10, 2006 15:49 UTC (Sun) by khim (subscriber, #9252) [Link] (1 responses)

I just don't see how not reimplementing from scratch for Mac can be used as a negative point.

Easy: if you have sensible format then you don't need full windows importer to port it. You can grab separate already-written pieces (MathML, PNG, SVG, etc), write glue code from scratch - and get full importer.

KWord does not use ODF-importer from OOo, for example. Even if both OOo and KWord use libpng to import png images.

But when your format is so complex and when its parts are so intertwined that you only have all-or-nothing choice - then yes, the only sensible way is to port existing importer...

I'n not saying that MacWord team did bad - I'm saying that if what they did is good... then OpenXML must be pretty awful for it to be good decision.

Two links people might find interesting in this context

Posted Dec 10, 2006 16:32 UTC (Sun) by fergal (guest, #602) [Link]

write glue code from scratch - and get full importer.

Why would they even write the glue code from scratch? As the blog posting points out "There is a lot of commonality between the in-memory data model for Win32 Office and Mac Office, since they share a lot of the same code". Writing from scratch should be the last resort. That would be like Win-OpenOffice and Linux-OpenOffice having independent code for importing the same documents.

The comparison between KWord and OOo is bogus. They don't come from the same codebase to start with so there would be very little that they could reuse. Let's imagine they somehow did share the ODF importer code, would that imply that ODF was so horrible and complex the the KWord team had to reuse the OOO code?

I haven't read the spec for OpenXML but I believe everyone when they say it stinks. There are apparently lots of reasons. That MS are resusing code between Win-Word to Mac-Word is not one of them.