|
|
|
Sam managed to compile FreeM-0.3.0 on Win NT using the LCC-WIN32 compiler. But
|
|
|
|
Sam hacked it somewhat (#ifdef LCC_WIN32 and #endif and some small code changes
|
|
|
|
as well).
|
|
|
|
|
|
|
|
Sam encountered the following MAJOR problems in the port:
|
|
|
|
|
|
|
|
1. The total lack of termios support on the Win32 platform.
|
|
|
|
|
|
|
|
Sam simply made all of that conditional on LCC_WIN32.
|
|
|
|
It affects all terminal IO and such matters.
|
|
|
|
|
|
|
|
2. The poor support for signals (probably in the LCC compiler's library...
|
|
|
|
but could be intrinsic)
|
|
|
|
|
|
|
|
Again, using the OLDUNIX support it worked. Sort of.
|
|
|
|
And Sam did change a lot of signal to the 'generic' SIGBRK.
|
|
|
|
|
|
|
|
3. The lack of forking.
|
|
|
|
|
|
|
|
Since the library on the Win32 lacks a fork() call it was simply disallowed.
|
|
|
|
The other calls as well were changed to never fork,
|
|
|
|
or never expect signals from other jobs. Not nice.
|
|
|
|
|
|
|
|
4. All direct calls to Unix / Linux routines had to be changed.
|
|
|
|
|
|
|
|
|
|
|
|
And it compiled. And started. And all normal M code works!
|
|
|
|
BUT all escape sequences show up (as [ABC and what have you).
|
|
|
|
|
|
|
|
SO, I propose some changes to the guidelines for the final version PROVIDED
|
|
|
|
PORTABILITY is an issue:
|
|
|
|
|
|
|
|
1. NO RELIANCE on any particalur terminal should be made (possible UNLESS
|
|
|
|
explicitly called for in an OPEN statement). In other words, the system should
|
|
|
|
NEVER on its own output escape sequences or try to use colours; all the fancy
|
|
|
|
terminal handling in service.c should be deleted or moved to an OPEN statement
|
|
|
|
or something. |
