Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cdt-dev] Mac CDI Patches (Was MinGW gdb)

On Feb 4, 2010, at 2:07 PM, Marc Khouzam wrote:

-----Original Message-----
From: cdt-dev-bounces@xxxxxxxxxxx
[mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Michael Jackson
Sent: February-04-10 1:42 PM
To: CDT General developers list.
Subject: Re: [cdt-dev] Mac CDI Patches (Was MinGW gdb)

I'd like to comment on "why" those CDI patches were pushed,
at least from my perspective since I seem to be the "vocal one".
  In an email with Marc-Andre Laperle he indicated that a
Patched CDI still works better than DSF-GDB on OS X but there
is more to it than simply that. I am all about moving away
from CDI but I _still_ need something that works at least as
good as CDI. So here were my own
fears: The Galileo Service Release is coming up "real soon"
and from my guessing (listening in on the CDT conference
calls) was that DSF- GDB was most likely NOT going to be
ready by the time Galileo Service release was released.
Looking forward the only "Mac" person that is really working
on getting DSF-GDB working is a student. Will Marc- Andre be
able to get all the necessary code patches, QA and other
stuff done in time for Helios? Plus twist someone's arm to
get all those patches reviewed, committed and tested? I'm the
"glass is half full" so I'm going to put a 50/50 bet
something goes wrong OS X Debugging for Helios is still
worthless. When is the next chance after that? November of
2010? I'm tired of waiting for it to be fixed.
  Patching CDI, however "dead" and deprecated CDI is, STILL
gives a BETTER debugging experience than current Galileo or
DSF-GDG can give.
The patches seem to be non-intrusive and well
compartmentalized which means the risk for other platforms is
low. Now that those patches are in the CDT 6.x CVS branch I
can have an "Official" CDT distribution with Debugging that
actually works. It is all about what I can get NOW and that
seems to be CDI.

I hear your pain and I can't argue against your approach.
You need to debug on Mac, and you will use whatever works.

Let's get DSF-GDB to work then, since it is so close to what
CDI offers.  You had written:

A fully patched CDI works better than DSF on Mac OS, at least,
from what I tested so this is what I use. Here's what works in
a fully patched CDI versus DSF-GDB :
- Properly display threads without refreshing (Bug 301720)

I'll look at this one since Marc-Andre kindly posted a patch.

- Debug in a path containing a space (bug 263689). To be fair,
it probably doesn't work with DSF-GDB on any platform. It would
also be hard to do a Mac OS-specific patch in DSF-GDB.

We could do a Mac-specific patch by using the service instead
of sending the MI command directly.  I'll put that comment in the
bug to help it along.

For my info, are spaces used a lot in paths on Mac?
I've never had to care about this problem on Linux.

- Inserting a breakpoint while GDB is running. This is also not
supported in DSF-GDB on any platform, see bug 242943.

I've always wanted to have this fixed.
I can look at that one too since I see a patch from Marc-Andre.

As a note though, there are more and more people knowledgeable
about DSF in the CDT community these days, and maybe someone
else wants to give one of these bugs a look?  I hate to think
I'm the reason these things are not being fixed, but I also have
a lot (_lot_) of other work to do.



If you take a look at an email I sent to the list on January 14, 2010 entitled "Re: [cdt-dev] Mac debug needs attention (patches are available!)", it has a nice summary of the "critical" bugs that really need the patches committed/tested. Marc-Andre is doing a really wonderful job with these patches. Thanks Marc-Andre.

And I understand we all have our "day jobs" in addition to the time we want to devote to CDT. I will not have any time until after April at which point I might be able to dedicate a few more cycles to the cause.

I don't think you are the "cause" of things not getting committed. What I think needs to happen is that there needs to be an advocate for OS X debugging issues on the CDT Dev team. Right now only community members are the voice and without having CVS commit rights, while that voice can be loud at times, that voice basically worthless. Lately the Mac CDT Community have gotten great support from Ken and Marc and I truly appreciate you taking time out of your busy schedules to review and commit the fixes.

Mike Jackson.

Back to the top