Chris Hanson [Wed, 24 Feb 1999 05:59:23 +0000 (05:59 +0000)]
Pass continuation's BLOCK-THREAD-EVENTS? member through the stack
parser. This isn't right, but unless we implement
WITH-THREAD-EVENTS-BLOCKED, there's no way to do better.
Chris Hanson [Tue, 23 Feb 1999 06:13:14 +0000 (06:13 +0000)]
Fix bug: when running something in a REPL buffer, Edwin would
completely ignore all input. It turned out that the "select"
mechanism was only looking at events for the console window, and not
for other windows, so it was blocking indefinitely.
Chris Hanson [Mon, 22 Feb 1999 05:05:10 +0000 (05:05 +0000)]
Fix bug: when switching between fvwm2 virtual desktops, Edwin wasn't
redisplaying immediately if it was busy doing something else. This
happened because it was not processing "expose" events until it
returned to the command level.
Chris Hanson [Thu, 18 Feb 1999 04:04:43 +0000 (04:04 +0000)]
* Add mechanism to encapsulate one port in another, and to build
wrappers around selected operations on the encapsulated port.
* Use new encapsulation mechanism to reimplement transcript facility
so that each transcript is associated with a particular REPL.
Previously the transcript was directly associated with the console
port.
This change is the goal of all of the port changes from this past
week. (I'm a little surprised at the depth of changes required.)
This has the side effect of increasing modularity, since the
transcript code is now concentrated in one file rather than being
integrated into the console port.
* Export procedure OUTPUT-PORT/FRESH-LINE to the global environment.
This was an oversight from previous changes.
Chris Hanson [Thu, 18 Feb 1999 03:54:37 +0000 (03:54 +0000)]
* Add mechanism to encapsulate one port in another, and to build
wrappers around selected operations on the encapsulated port.
* Use new encapsulation mechanism to reimplement transcript facility
so that each transcript is associated with a particular REPL.
Previously the transcript was directly associated with the console
port.
This change is the goal of all of the port changes from this past
week. (I'm a little surprised at the depth of changes required.)
This has the side effect of increasing modularity, since the
transcript code is now concentrated in one file rather than being
integrated into the console port.
* Export procedure OUTPUT-PORT/FRESH-LINE to the global environment.
This was an oversight from previous changes.
Chris Hanson [Tue, 16 Feb 1999 19:44:51 +0000 (19:44 +0000)]
Change the port implementation to have a type that holds the
operations on the port. This new implementation supports a crude form
of single inheritance.
Chris Hanson [Sun, 31 Jan 1999 20:48:23 +0000 (20:48 +0000)]
Change name: RUN-SYNCHRONOUS-PROCESS to RUN-SYNCHRONOUS-SUBPROCESS.
Change calling convention of RUN-SYNCHRONOUS-SUBPROCESS and
RUN-SHELL-COMMAND to accept keyword arguments, and eliminate
MAKE-SUBPROCESS-CONTEXT.
Chris Hanson [Sun, 31 Jan 1999 20:46:25 +0000 (20:46 +0000)]
Change name: RUN-SYNCHRONOUS-PROCESS to RUN-SYNCHRONOUS-SUBPROCESS.
Change calling convention of RUN-SYNCHRONOUS-SUBPROCESS and
RUN-SHELL-COMMAND to accept keyword arguments, and eliminate
MAKE-SUBPROCESS-CONTEXT.
Chris Hanson [Sun, 31 Jan 1999 20:43:21 +0000 (20:43 +0000)]
Change name: RUN-SYNCHRONOUS-PROCESS to RUN-SYNCHRONOUS-SUBPROCESS.
Change calling convention of RUN-SYNCHRONOUS-SUBPROCESS and
RUN-SHELL-COMMAND to accept keyword arguments, and eliminate
MAKE-SUBPROCESS-CONTEXT.
Chris Hanson [Fri, 29 Jan 1999 05:33:57 +0000 (05:33 +0000)]
Change prompt history yet again. Now the history is copied before
use, and the user's editing changes are remembered for each element in
the history. When the user exits the minibuffer, all of the changes
are discarded.
Also: eliminate HISTORY-DEFAULT option; HISTORY-INDEX now subsumes
that role.
Chris Hanson [Thu, 28 Jan 1999 04:00:18 +0000 (04:00 +0000)]
Revamp of prompting code. New design supports keyword arguments to
most prompting procedures, to support options in an extensible way.
The new keyword options are used to implement a general history
mechanism, like that previously implemented by repeat-complex-command
(which is now implemented using the new mechanism).
This edit has made incompatible changes to the calling conventions of
the following procedures: