Shortly before and at EclipseCon I received a lot of questions about
the effort to improve the multi-context debugging experience in
Ecilpse. If I had the time, I explained what the current situation of
this DD sub-group is:
This brings us to this point in time, where the multi-context effort is
stalled, but I get the sense that there is a lot of demand to re-start
this discussion. I personally don't have a lot of time to allocate to
this effort at the moment, and I think I need to have a clear idea of
what changes features are needed before I ask my management for time to
work on it. Therefore, I would like to re-start the multi-context
group in the way that I should have started it in the first place. As
a community that has a vested interest in this problem we need to
- About a year and a half ago, I prototyped an enhancement to the
platform debugger views to allow the user to create multiple instances
of views and to pin them to the active context (see bug 145625).
- At the meeting
in Toronto in January '07 I demonstrated my prototype and we had a
lively discussion about its benefits and short comings.
- We decided that the bug 145625 prototype was too radical as a
workflow change to introduce in Platform without considerable
discussion. It also was not a complete solution to the multi-context
workflow problem and there were some implementations issues with it.
- I promised to write up a proper proposal for the pin and clone
workflow changes in 3.4 time frame which could be properly evaluated
and potentially contributed to platform Debug.
- As I considered the multi-context workflow issue, I realized that
there are a lot of issues to consider and I tried to write a proposal
that would try to address these issues (in two parts). I wrote the
first part of the proposal and posted it to wiki.
- I received a lot of excellent feadback on the first document, and
it quickly became clear that I tried to cut too many corners in writing
the proposal because I omitted use cases, mixed functional description
with design and didn't have enough content.
- I never got to writing the second part of the proposal as my
commitments for DSF and Wind River commercial work took priority. I
put the effort on hold until after ganymede release effort is under
- Define the use cases, and identify the use cases which are most
- Identify the problems with executing the use cases.
- Come up with ideas to fix the problems and improve the
multi-context debugging user experiecene.
In the past discussions about multi-context debugging, we used a
combination of wiki and the mailing list. This was somewhat confusing
and required some cut-and-paste action just to make sure that the edits
were seen. To avoid this, I would like to use the http://wiki.eclipse.org/DSDP/DD/MultiContext
Wiki page as the center of this discussion. To facilitate this I ask
that anyone that wishes to participate in this discussion add this page
to their wiki
watch list and configure your wiki preferences to email yourself
when the watchlist is updated. This should hopefully minimize the need
to use the mailing list just to tell everyone that the wiki page was
updated. The last section on the page is a "Comments" section, however
I hope that we can produce this document as a community and I hope that
everyone that wishes to participate in it will just edit the document
I started on this process tonight by adding some general use cases for
multi-threaded debugging. As I have time I will contribute more
content, but I hope to see others contribute to it as well. I also
welcome any feedback and suggestions as far as this process itself.