| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
* fixes at least <FS>, which has been broken since b5325e00c402ec18034da4b4a4aaaefa87bb1fef.
* but probably many other commands as well in the DOS version when working with
"very large" documents
* now compiles cleanly with -Wsign-compare
|
|
|
|
In particular, this will work on MS-DOS, albeit inefficiently.
You also cannot generically redirect stderr on COMMAND.COM.
|
|
* This allows handling files >64kb even on 16-bit DOS.
* A few flags fields could be squashed into bitfields.
* The buffer overview (0EB) has been adapted for DOS.
We can only show the total amount of allocated
memory (as by tecmem.c) for the time being.
Unfortunately, we don't have a total amount of available
memory since _memavl() is for small data models and
the sbrk() apparently doesn't grow.
|
|
It can be cross-compiled or compiled natively on 32-bit DOS
with OpenWatcom C v1.9.
Compiling on an 8086 might be possible later on - but we would
have to add support for some ancient ANSI C compiler.
|
|
Also avoid old-school function declarations without parameters (`void foo();`).
This is unnecessarily loosing type safety.
|
|
|
|
|
|
* undo token for DOT change, necessary if DOT was in EC argument range
* use read/buff_insert loop to insert new text so we don't have to tweak DOT to be able to use buff_readfd and can do pipe-specific stuff
|
|
unidirectional pipe with no arguments. with one/two arguments,
the buffer content is piped into the process and replaced with
its output.
* fixed error handling, now the child's exit status matters
if everything else is OK, as well
* colon modifier has standard behaviour of returning the command
status
|
|
etags/ctags functions
|
|
* adapted Doxygen config file
|
|
to Doxygen format and cleaned up manually
|
|
* makes special handling of Linux unnecessary
* enables the (undocumented) :EC mode where the entire buffer content is written to the process (two pipes are created, both stdin and stdout/stderr are redirected) on Linux
|
|
|
|
* errno was evaluated even though it was in an undefined state
* error_message()s written by cmd_writebak() were overwritten by cmd_write()
|
|
* now files can be saved, but the current heuristics of creating a .OLD once and afterwards updating a .BAK might be nonsense
* behaviour without HAVE_LONG_FILE_NAMES should work but does not make sense IMHO: will remove support for that case anyways
|
|
|
|
|
|
|
