As I write this, it’s the third week of June, and I’m on a consulting gig in Boston doing some cool high-performance Linux stuff. Unfortunately though, I’m stuck fifteen miles outside the city, in the suburb of Woburn, MA, at an extended stay efficiency suites hotel (albeit a nice one) because I couldn’t get anything closer without bursting a blood vessel of a corporate bean counter. So, naturally the morning commute to downtown Boston is brutal.
Why the long commute? Well, as it turns out, this week is Tech Ed, the big Microsoft education and training conference, where the software behemoth brain dumps on developers and systems integrators, sharing what Redmondware is coming in the the next year. And what do you know? Oe of the big themes that’s apparently come up this week is interoperability of Microsoft products with Open Source.
I’ve been saying for a long, long time that Microsoft should do pursue interoperability with Open Source, and I’ve gone so far to say that Gates’s crew should actually Open Source components of Windows NT, Windows 2000, or large parts of the operating systems wholesale, but so far, my cries have been fallen on deaf ears. At this point, I’m hoping that Microsoft at least takes some baby steps, because any concession from the software giant would be a good one. Interoperability between Windows infrastructures and Open Source operating systems and applications certainly would be a good thing.
I can think of a number of efforts they could start right away that would make all of our lives in the IT consulting field a hell of a lot easier:
*Microsoft should join the SAMBA project and open source or at least publish the API’s and complete specifications for Active Directory’s modifications to LDAP and Kerberos. Moeover, the company should open its latest CIFS implementation, so PAM modules for Linux can natively login to Windows 2000 Active Directory domains without requiring the assistance of a rocket scientist or proprietary reverse-engineered Active Directory integration products, such as Centrify, Vintela, Centeris, and PADDL.
*Micorosft should publish the complete specifications to the NTFS file system, so Linux users can finally dual-boot and read and write NTFS partitions without fear. Sure, it works now with the reverse engineered Open Source stuff that’s out there, but the real McCoy would be so much better. With an encrypted NTFS due out with Windows Vista in 2007, the current, independent Open Source drivers will only work for so long.
*How about opening the source to the INTERIX stack (the POSIX pieces of Microsoft Services for Unix), so we can build a complete Linux Standard Base- compliant POSIX subsystem for Windows Vista and Windows Server to facilitate cross platform development of Open Source software on Windows. The current MSFU 3.5 POSIX implementation sucks dead gophers thru rusty tailpipes, and it’s practically worthless unless you compile a ton of software on it. MSFU 3.5 is already free of charge, it wouldn’t be a huge step to Open Source it completely.
*Open Source the Windows NFS client and server in MSFU 3.5, because it could could use some serious optimization and repair — as it stands now, it’s buggy as hell.
Create a Microsoft equivalent of SourceForge/Freshmeat and a Open Source version of MSDN.com
for the express purpose of developing Open Source software on Windows. Have a centralized download area for getting key Windows ports of Open Source projects, such as GIMP, GAIM, Audacity, Apache, MySQL, CLAMAV,
and so on. (Note to Microsoft: Have a look at http://osswin.sourceforge.net/
. I’m quite sure you can do a better job.)
Sure, this list has some some pretty tall orders. But if Microsoft really wants to build some bridges, it has to provide some serious steel for the bridge builders.
And for crying out loud, someone get me a closer hotel to Chelsea than Woburn!