clayborg added a comment.
Sounds good. This code is very tricky and we need to find a way to control this
so that these issues never happen. When we do this, it will hopefully fix these
kinds of issues for everyone. Thanks for taking the extra time to look into it
and at least you have made the
ted added a comment.
Adrian and I worked out what is going on:
1. Run starts the launch and writes a prompt
2. State change fires, prints out “launching” through async, overwriting the
prompt from 1) and writing a new prompt
3. CommandObjectProcessLaunch::Execute appends “launch” to the result.
amccarth added a comment.
This is better. In the same scenario, I see just one extra prompt instead of
three extras. Maybe that one happens through a different code path.
> (lldb) br set -l 22
> Breakpoint 1: where = a.exe`main + 20 at fizzbuzz.cpp:22, address =
> 0x0040e074
> (lldb) run
ted updated this revision to Diff 51462.
ted added a comment.
Updated to change cursor position before async output so old prompt is
overwritten.
Adrian, please test with your setup and let me know if it solves the issues you
see.
http://reviews.llvm.org/D18335
Files:
source/Core/IOHandler
ted added a comment.
Without the patch I'm seeing the original prompt printed after the command, in
a line like this:
(lldb) Process 1 stopped
and no prompt after the async output from the stop.
With the patch I see the above prompt, and a prompt after the async output. I
think with the patch
clayborg added a comment.
Seems like it would be nice to get the prompt clearing working...
http://reviews.llvm.org/D18335
___
lldb-commits mailing list
lldb-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
amccarth added a comment.
Here's a snippet of output with this fix patched in on Windows. Note the extra
(lldb) prompts on the lines beginning with "Process 10672..."
(lldb) br set -l 22
Breakpoint 1: where = a.exe`main + 20 at fizzbuzz.cpp:22, address = 0x0040e074
(lldb) run
Process 10672 laun
clayborg added a comment.
this could still be the same bug. So if this doesn't fix things for you on
windows, then this bug isn't quite fixed. Ted, are you seeing the same thing?
amccarth, can you attach some output to see what you are seeing?
http://reviews.llvm.org/D18335
clayborg added a comment.
It would be nice to clear the line though. What _should_ happen is the "(lldb)
" prompt is cleared by clearing text to the start of where the prompt was
printed, then display the async text and refresh the prompt. So this fix is a
bit hacky in that it isn't trying to s
amccarth added a comment.
I patched this change in, and now I'm getting extra prompts on Windows rather
than getting the too-late prompt sooner. So is that a separate bug?
http://reviews.llvm.org/D18335
___
lldb-commits mailing list
lldb-commits@l
clayborg added a comment.
No, this fixed the problem of printing the "(lldb) " prompt first, then being
asked to print async text, then not refreshing the prompt. Not to say there
won't still be a race, but this particular problem is what we are fixing.
http://reviews.llvm.org/D18335
__
amccarth added a subscriber: amccarth.
amccarth added a comment.
This just seems to insert an extra prompt rather than solving the problem with
the prompt being printed too late.
http://reviews.llvm.org/D18335
___
lldb-commits mailing list
lldb-com
clayborg accepted this revision.
clayborg added a comment.
This revision is now accepted and ready to land.
Looks fine.
http://reviews.llvm.org/D18335
___
lldb-commits mailing list
lldb-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/li
ted created this revision.
ted added reviewers: clayborg, zturner.
ted added a subscriber: lldb-commits.
On Windows (and possibly other hosts with LLDB_DISABLE_LIBEDIT defined), the
(lldb) prompt won't print after async output, like from a breakpoint hit or a
step. This patch forces the prompt t
14 matches
Mail list logo
