Josh Ledgard On MS's Future Open Source Efforts

prostoalex writes "Josh Ledgard from Microsoft, the developer responsible for open-sourcing WiX and WTL, is looking for opinions on what Microsoft should do next in regards to the open source movement that he himself established within the company. "Would you have interest in working on these types of projects with Microsoft? If not, what could entice you? If so, what would be your motivation?", asks Josh." Update: 08/24 19:04 GMT by T : As Ledgard writes on his site, "I am NOT the person responsible for the WIX/WTL projects. I cite them as examples and am working with people who where responsible for those projects to enable more of the same for the groups I work in." Sorry for the misattribution!

Comments

Re:Future Open Source efforts?

[MoonFog]

Yes, they even won an award for that one.

Re:Office Specs..

[gagravarr]

They do. You can license them. Trouble is, once you have, you're bound by quite a number of restrictions One of the POI developers (Jakarta Word/Excel project) recently had to stop, as he'd changed employers to someone who'd licensed the specs: leaving email What would be nice is getting those docs without all the heavy licensing

Re:Office..

[sql*kitten]

...or at least give .doc format(s) specifications (if they even exist)

MS Office files are actually miniature filesystems in their own right, containing serialized objects that are the in-memory representation of a document. There is no intermediate step between the objects that are the document in-memory that translates it into ".doc format" on the disk - it's just the object itself.

So:

1. The "standard" is the class that represents the document (i.e. the code of it + the generic object serialization code)
   
2. The "standard" changes between versions because the document classes change as new features are added. There is no deliberate policy to "break" things between versions, it is just a side effect.
   
3. It is easy to use COM to instantiate Word from your own code and manipulate documents throught the API, so ".doc format" is fully accessible and reusable from your own code, just as it would be if it was "open source".

Re:Just as if it were open source...

[IamTheRealMike]

Open source means the source is open. We can all see it. This means we would be able to see the code which creates the .doc file from start to finish. This also means we would be able to write our own library to read and write .doc files completely with our own code on any platform because we can see exactly how the original works.

Hahaha, I guess you've never read the DOC specs. I read a very old version of them that was distributed/leaked one time on MSDN many years ago - you can get it from the WvWare homepage.

Basically the Word specifications are nearly incomprehensible to human beings, let alone the code. Given that the specs actually partly consist of copy/pasted code I think even if Word were open sourced we'd only be a bit further in having OpenOffice and the like able to read them.

The Word developers use structures with members names like fcPlcffldHdrTxbx. It's hungarian taken to the logical extreme. Here's a line from one of the struct definitions:

FTC rgftcStandardChpStsh[3]; // ftc used by StandardChpStsh for this document

Helpful, isn't it? I'm convinced the Word team are actually robots from the 23rd century, no people could have produced this.

Mind that the specs appear to be internal documentation yet are still incomplete!

Basically it would not surprise me if the internals of Word are like the internals of Windows: the people who truly understood them left or retired years ago leaving a mess of interlocking systems and hacks that nobody can fully comprehend.

Re:OpenSource IE

[jonwil]

IE is integrated into the OS in the sense that if you remove it, a fair few things (like lots of functionality in Explorer, also lots of parts of Windows Help and HTMLHelp) would stop working.

However, most of the IE related code is in several seperate files, specifically:
hlink.dll (HyperLink Library)
iexplore.exe (main program exe)
imgutil.dll (Image Decoder API)
mshtml.dll (HTML rendering engine)
jscript.dll (JScript runtime)
shdocvw.dll (Internet Explorer core)
wldap32.dll (Win32 LDAP implementation)
msxml.dll (XML parser)
pngfilt.dll (PNG library)
and probobly a few others as well.

Microsoft would have nothing to loose and probobly a fair bit to gain if they were to Open Source those bits that are "IE" (and not the other bits connected to, related to or used by it e.g. Windows Media, Shell, Windows Core, Outlook, OLE, COM, Common Controls etc)

Especially if they then took the best "user changes" and put them into the official tree and future IE updates.

Re:Future Open Source efforts?

[zerblat]

When Microsoft bought Entropic, they released the source code for ESPS under a BSD-type license. ESPS is a very cool speech processing toolkit that used to be heavily used in research. Unfortunately, AFAIK the code doesn't compile out of the box. Some of the code from ESPS is however used in Snack/Wavesurfer.