[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
[dsdp-dd-dev] Re: Refined Disassembly Requirement
|
OK, now I think I understand, let me restate:
1) Disassembly alone (and by extension address-based disassembly) cannot
be shown in the editor, because of implementation difficulties.
2) Source-based disassembly should be implemented by having an editor
API so that disassembly can be interjected in between source line within
any editor.
For #1, it is possible to show address-based disassembly in editor,
which is what we do in our product. It's not a perfect user experience,
but scrolling an essentially infinite range is never perfect. I also
don't know the implementation details, they were done by Toni and he
could provide more details. So I wonder if anyone thinks that if we
could have the address-based disassembly in the editor, that it would be
a better user experience to display disassembly in the editor space,
rather than (or in addition to) having it in a view.
For #2. I actually do like the idea very much, I can see one potential
usability problem with it though. If user is debugging two instances of
the same process at the same time, he would normally have both IP
annotations painted in the same editor at the same time. If the user
then decided to switch to mixed mode, which disassembly should be shown
in the editor? I'm sure there are solutions to this, but they could be
pretty messy.
Cheers
Pawel
Chuong, Patrick wrote:
Pawel,
In the long run, we do need to have a disassembly view. The editor is
tied to source file, is this correct? You wouldn't be able to show
disassembly if there is no source, and it is not possible to show
disassembly for the entire program address space.
Mikhail's note means that we should make the disassembly component
contribute the assembly text for each source line in the editor, and
this becomes the mixed mode. Hence, we might not need to have the
disassembly view to show mixed mode. I believe this is much better than
having the disassembly view to show mixed mode, you get all the syntax
highlighting for free.
-patrick
-----Original Message-----
From: Pawel Piech [mailto:pawel.piech@xxxxxxxxxxxxx]
Sent: Thursday, June 22, 2006 4:36 PM
To: Mikhail Khodjaiants
Cc: Chuong, Patrick; John Cortell; Samantha Chan; Spear, Aaron; Duane
Ellis; Ewa Matejska; Swiezawski, Martin
Subject: Re: Refined Disassembly Requirement
I'm more confused still then. Your design note in disassembly page
says: "Showing disassembly in the editor is the preferred solution, and
it should be implemented as a contribution to an editor view an
extension point or something similar (Mikhail Khodjaiants).". I took
that to mean that you do want to have the disassembly shown within an
editor part, did I misunderstand?
Thanks
Pawel
Mikhail Khodjaiants wrote:
Pawel,
The short answer is simplicity and similarity to other debug views.
And the disassembly presentation in the early CDT versions was
implemented as an IEditorPart object, but it was a viewer anyway. We
have never had a disassembly presentation embedded into the C editor
and I still think it is not a good idea.
We can easily improve the current implementation by adding the
features mentioned in the Patrick's document. Using "clone and pin"
approach we will get the same functionality we had with disassembly
"editors".
Mikhail
----- Original Message ----- From: "Pawel Piech"
<pawel.piech@xxxxxxxxxxxxx>
To: "Mikhail Khodjaiants" <mikhailk@xxxxxxx>
Cc: "Chuong, Patrick" <pchuong@xxxxxx>; "John Cortell"
<john.cortell@xxxxxxxxxxxxx>; "Samantha Chan" <chanskw@xxxxxxxxxx>;
"Spear, Aaron" <aaron_spear@xxxxxxxxxx>; "Duane Ellis"
<duane_ellis@xxxxxxxxxxxx>; "Ewa Matejska"
<Ewa.Matejska@xxxxxxxxxxxxxx>; "Swiezawski, Martin"
<mswiezawski@xxxxxx>
Sent: Thursday, June 22, 2006 1:23 PM
Subject: Re: Refined Disassembly Requirement
I feel like I'm missing a bit of history here. Does anyone know what
was
the reasoning behind switching the disassembly editor to be a
disassembly view in CDT?
Thanks
Pawel
Chuong, Patrick wrote:
Hi all,
I am sending this out offline instead of broadcasting it to the DSDP
mailing list, and I have added everyone's name that has responded to
my original email to the DSDP mailing list. I have refined the
disassembly requirement and reorganized it for clarity. If it is
possible I would like to have further feedback for missing
requirements or if there is any requirement that you would like to
discuss. You can go to the Wiki website to review the changes.
Please use the existing disassembly requirement thread to continue
the discussion.
Thanks,
Patrick
