Re: LYNX-DEV My planned version

[Michael Sokolov]

   Leonid Pauzner <address@hidden> wrote:
> I think we may understand each other.
   
   Thanks!
   
> Yes, Michael Sokolov can made DOSSOCK/MMM386/LYNX386 [...]
   
   MMM386 is not a prerequisite for my work on Lynx. The reference
implementation of True DPMI is the Protected Mode DOS component of Windows
v3.xx, and that's probably what I will use for the development and testing
of my version of Lynx. It may very well end up being released before
MMM386. In this case as soon as MMM386 is released, Lynx will run under it
out of the box, as it will use only the implementation-independent parts of
the interface. Hey, that's the whole point of standards :-).
   
> From the other hand, when (and if) it will be done
> we may incorporate his DOS code to 2.8+ because it is a good idea to have
> port in progress so later additions/improvements to
> HTML/HTTP/etc. handling will be incorporated automatically
   
   What do you want to do with my code is entirely up to you. It will be
free software.
   
> (unless Michael plan maintain such lynx issues himself,
> I think not, sure).
   
   My version of Lynx WILL be up-to-date with the improvements in the Lynx
"matter". One of the following will happen:
   
   1. Fote will keep his personal code up-to-date. When I finish my DOS
support code, he will incorporate it in his personal version. In this case
I will be done.
   
   2. Fote will keep his personal code up-to-date, but won't accept my DOS
support code. In this case I will track his code and keep merging it with
mine.
   
   3. Fote will be satisfied with v2.7.2 as it is now and freeze it (or
stop using Lynx himself). In this case I will take over all of his
responsibilities, including tracking others' work on Lynx, separating it
into useful contributions and loose cannons, and including the former into
my codeset with proper clean-up and style adjustment.
   
> Thanks, Michail, for your explanation of True DPMI in Windows etc.
   
   If you want a real explanation of this, read Andrew Schulman's
_Unauthorized Windows 95_. All of its ~600 pages deal with this very
subject. My explanation is a feeble attempt to summarize Andrew Schulman's.
As you can imagine, it's anything but easy to summarize ~600 pages in three
paragraphs.
   
> Another (minore) difference may be from long filenames with spaces etc.
   
   This is due to a slight change in the Protected Mode DOS component
between Windows v3.xx and Windows 95. Several new INT 21h API functions
have been added (AH=71h most notably) that allow DOS apps to use LFNs (long
filenames). (Yes, LFNs are due to a change on the DOS side of the fence,
not on the Windows side. Win32 apps automatically get LFNs only because
Win32 API calls are implemented in terms of the new LFN-based INT 21h
calls. For example, CreateFile() becomes INT 21h/AX=716Ch. In general,
Win32 is implemented on top of and in terms of DOS32. That's why there is
absolutely no need for a Win32 version of Lynx: a DOS32 one is a much
better idea.) All of this is described in gory details in _Unauthorized
Windows 95_.
   
   I probably won't use the LFN functions in my version of Lynx. After all,
it's a DOS app (albeit a 32-bit protected-mode one), and no one requires
DOS apps to use the LFN functions, even if they are available. Win95 users
can run it just as they would run any other DOS software (16-bit or 32-bit,
real or protected mode).
   
   Sincerely,
   Michael Sokolov
   Phone: 440-449-0299
   ARPA Internet SMTP mail: address@hidden
   
   P.S. I have invented the terms like "Protected Mode DOS" and "DOS32"
myself in order to illustrate the point that the work that Windows usually
gets credit for is really done by a new form of DOS. The correct name for
the latter is DPMI (DOS Protected Mode Interface).