Re: [Flightgear-devel] SGPropertyNode troubles

2007-02-17 Thread Melchior FRANZ
* Csaba Halász -- Friday 16 February 2007: > Preliminary proof-of-concept patch is attached [...] Committed. For those who have become a bit uncomfortable in the light of recent changes to the SGPropertyNode class -- FlightGear's nervous system: I think it's done now. Three old bugs fixed, and no

Re: [Flightgear-devel] SGPropertyNode troubles

2007-02-16 Thread Maik Justus
Hi Melchior, as discussed in IRC, I agree with Jester, that he has found a bug in my property-cache-patch. His patch looks very reasonable to me and I vote for committing it to cvs. Thanks, Maik Melchior FRANZ schrieb am 16.02.2007 10:32: > * Csaba Halász -- Friday 16 February 2007: > >> On

Re: [Flightgear-devel] SGPropertyNode troubles

2007-02-16 Thread Maik Justus
Hi Melchior, Melchior FRANZ schrieb am 16.02.2007 12:34: > OK, I found one old bug, one new bug, and one new bug in dead > code (where it doesn't do any harm, but bug is bug). > yes, there was a bug in the dead code, but it only wasted some cpu cycles. But removing _data[index]->get_

Re: [Flightgear-devel] SGPropertyNode troubles

2007-02-16 Thread Csaba Halász
On 2/16/07, Melchior FRANZ <[EMAIL PROTECTED]> wrote: > > one new bug in dead > code (where it doesn't do any harm, but bug is bug). If you mean the hash_table::erase(const char*) function doing funny things, then I've found that too. > As I couldn't reproduce your problem before, I can also not

Re: [Flightgear-devel] SGPropertyNode troubles

2007-02-16 Thread Melchior FRANZ
OK, I found one old bug, one new bug, and one new bug in dead code (where it doesn't do any harm, but bug is bug). As I couldn't reproduce your problem before, I can also not say if this fixes it. Let's talk about that on IRC, where all the important things happen. ;-) m. ---

Re: [Flightgear-devel] SGPropertyNode troubles

2007-02-16 Thread Melchior FRANZ
* Csaba Halász -- Friday 16 February 2007: > On IRC ndim and I have been investigating a segfault (backtrace here: > http://flightgear.lauft.net/yet-another-segfault-in-simgear.bt) Eew. I'll look into it later today, and discuss the problem with Maik. Haven't seen this crash yet (happens on exit a

[Flightgear-devel] SGPropertyNode troubles

2007-02-15 Thread Csaba Halász
Hello everybody! On IRC ndim and I have been investigating a segfault (backtrace here: http://flightgear.lauft.net/yet-another-segfault-in-simgear.bt) The culprit seems to be the path cache in simgear/props/props.cxx. Here is a little overview of what we thought: Jester: could it be that _linke