Custom Query (1030 matches)

The behavior of the return key in the listener in the latest release seems to have changed in a very annoying way. The previous behavior was:

1. If the cursor was on the last line of the listener, and there was

a complete sexpr on that line, that sexpr would be evaluated.

2. If the cursor was on the last line of the listener and there was

not a complete sexpr on that line, a newline would inserted.

3. If the cursor was anywhere else in the buffer, the sexpr to the

left of the cursor would replace the last line in the buffer. (IMHO, this was not the correct behavior. The correct behavior is what Fred used to do: append the sexpr to the left of the cursor to the last line. But that's another issue.)

The new behavior, as best I can make out, is:

1. If the cursor is at the end of the buffer (not merely on the last

line) and there is a complete sexpr to the left of the cursor then the sexpr is evaluated. This is as it should be. However...

2. If the cursor is on a line other than the last, then the sexpr on

that line is copied to the last line AND it is evaluated. This is badly broken IMHO because there is no opportunity to edit the line. Now to re-use a previous line of input with changes you have to select, copy, click, and paste. Very annoying. Worse...

3. If the cursor is on the last line but not at the end of the line,

then a newline is inserted. In addition, if there was a complete sexpr on the last line, it is evaluated. However, the cursor does not drop down to the new last line. It stays where it is. This is just b0rken. It's particularly annoying because there's a bug in the listener scrolling code so that if you do this at the bottom of a window, you get output that you don't see unless you manually scroll the window down.

I'm rating this "major" because it's really messing me up, especially the last item.

To reproduce:

1. build and launch the IDE
2. Press the ESC key
3. Press any other key

Result:

Error: value NIL is not of the expected type STRUCTURE.

While executing: (:INTERNAL GUI::keyDown:|), in process Initial(0).

Type :POP to abort, :R for a list of available restarts. Type :? for other options.

1 >

After exiting from the breakloop, the IDE remains confused, expecting a keydown and breaking on any it receives.

After obtaining a copy of Clozure CL 1.11 via:

$ svn co ​http://svn.clozure.com/publicsvn/openmcl/release/1.11/linuxx86/ccl

using my Raspberry Pi 2 B, the armcl runs, but it appears to have the "… lisp-kernel binary that is included in the Subversion repository uses the soft-float ABI." I would like to use the hard-float ABI so have attempted to recompile the code. Following the ideas found at

http://trac.clozure.com/ccl/wiki/PlatformNotes#LinuxARM

I changed directory and did the make clean && make step. I have repeated these steps numerous times making a number of changes over time. But the out-of-the-box code itself, with NO CHANGES, compiles with this warning:

cc -include ../platform-linuxarm.h -c ../albt.c -DLINUX -DARM -D_REENTRANT -D_GNU_SOURCE -DSVN_REVISION="16704" -g -O2 -marm -march=armv6 -Wno-format -o albt.o ../albt.c: In function ‘walk_stack_frames’: ../albt.c:95:14: warning: assignment from incompatible pointer type

next = (LispObj *)header;

and results in a segmentation fault. I have commented this line of code out to remove the error and remade this. The build looked good, but as it always does, it results in:

# ./armcl

Segmentation fault

No matter what changes I make, such as changing armV6 to armV7-a in the Makefile, commenting out warning lines, and even including none at all, I get this segmentation fault. It appears the CCL cannot be compiled on the Raspberry Pi 2 B even with no changes to the code downloaded via svn. Yet, it runs in soft-float mode directly after download, but I cannot rebuild it in soft or hard float mode.