Improving variables view

jbb jbb at kdevelop.org
Wed Jun 8 11:20:07 UTC 2005


On Wed, 08 Jun 2005 05:48, Alexander Neundorf wrote:
> Hi,
>
> On Tuesday 07 June 2005 10:23, jbb wrote:
> > On Tue, 07 Jun 2005 04:04, Alexander Neundorf wrote:
> > > Hi Vladimir,
> >
> > [snip]
> >
> > > Since it doesn't support the machine interface, it isn't of much use
> > > for me without some hacks. If debugging over the serial line gdb output
> > > and applicatio output come intermixed and kdevelop considered the
> > > application output as gdb termination message :-/
> >
> > Hmm, that shouldn't happen.
> > You see, I change the gdb prompt on issuing a command and change it back
> > again when the command is finished. This "brackets" the output from the
> > commands I send to gdb. All output from the program should be outside
> > these areas. Although where is goes, well, that's interesting. Can't
> > recall ...
>
> It arrives in GDBController::parseLine(), I had a problem with the match
> for "Cann", since this appeared also in my debug outpout:
>
> if (strncmp(buf, "Cann", 4) == 0)
> {
>    ...
>    actOnProgramPause(QString(buf))
> }
>
> And then kdevelop considered my debugging session paused :-/

hehe - such a wonderful piece of code. It's history is long and infamous. 
That's a bit of leftover code, in some ways, from when I tried to optimise 
the strings searches by converting the 1st 4 chars to a number. Hence all the 
cmp with 4 chars. Caused segfaults on some platforms so was changed back to 
the compares you see to today.

It's trying to filter out lines that fall out of gdb we didn't ask for. This 
piece of history and it's associated problems would be consigned to the bit 
bucket if MI was used. It could also be reworked easily to avoid your 
paticular pitfall. Try the patch - it's untested... the same sort of changes 
could be done throughout this method.

jbb
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gdbcontroller.diff
Type: text/x-diff
Size: 3811 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20050608/1de48a36/attachment.bin>


More information about the KDevelop-devel mailing list