home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!decwrl!hal.com!olivea!charnel!sifon!newsflash.concordia.ca!nstn.ns.ca!cs.dal.ca!iisat!mkseast!dale
- From: dale@mkseast.uucp (Dale Gass)
- Newsgroups: comp.os.ms-windows.programmer.win32
- Subject: NT Subsystems...
- Message-ID: <1993Jan26.211925.8015@mkseast.uucp>
- Date: 26 Jan 93 21:19:25 GMT
- Organization: Mortice Kern Systems, Atlantic Canada Branch
- Lines: 48
-
- Just got "Inside Windows NT", and am starting to understand the NT design
- a bit better... I do have some concerns, though...
-
- I can see the elegance in having every user process be a client of a
- subsystem, which is implemented in terms of NT kernel level calls.
- However, these NT kernel level calls are apparently not accessible
- from user level applications, only from the subsystems. If the subsystem
- decides to provide an API to grant access to the functionality provided
- by the kernel level API's, fine. If it doesn't, there's no way you can
- achieve that functionality within the subsystem.
-
- This strikes me as a serious disadvantage to this design approach.
- I am at the mercy of Microsoft's subsystem implementations in determining
- the functionality I can get out of NT. They don't provide any full-screen
- control from the Posix subsystem, so all my Posix applications are doomed
- to be pathetic interactively... This wouldn't be so bad if there were some
- way for the developer to work around it, but there doesn't appear to be
- any way to do so, short of re-implementing the subsystem from scratch.
-
- Also, I have seen no evidence of a "Subsystem Developers Toolkit", which
- would allow me to write my own subsystem to meet my all of my needs.
- I guess these kernel level calls are the undocumented ones which Microsoft
- doesn't document, because they are "subject to change"...
-
- What I would *love* to see would be either the availability of subsystem
- source code (not likely) so a developer could extend it (which would
- require all my customers to have the new subsystem, ugh!), or even better,
- add some hook for developers to *extend* the API's available from a
- subsystem.
-
- With the latter approach, for example, I could implement a full curses
- by adding the appropriate API's, written in terms of the kernel level
- API's. This "subsystem extension" (most likely a DLL) could allow adding
- any functionality to any subsystem, in whatever manner is most appropriate
- to the developer.
-
- Without many of the common extensions found on Posix compliant system
- (such as curses, terminfo, etc.), the Posix subsystem is more or less
- useless. It may meet the U.S. government FIPS, but no applications of any
- complexity could be made to work under it, so it will not be a Posix system
- of choice. I don't necessarily expect Microsoft to provide these extensions
- themselves, but I do expect them to make it possible for me to provide them,
- if I wish. Currently, they don't.
-
- -dale
- --
- Dale Gass, Mortice Kern Systems, Atlantic Canada Branch
- Business: dale@east.mks.com, Pleasure: dale@mkseast.uucp
-