Re: [Flightgear-devel] Built-in Svn client code crashing

2013-10-05 Thread Markus Wanner
James,

On 10/04/2013 08:43 PM, James Turner wrote:
 I haven't taken a copy of the libsubversion and
 forked/edited/trimmed it - it's completely unrelated code. 

Aha, I wasn't aware of that. Sounds like a different story, then.

 Does that still cause problems under to policies described above?

I don't think so. Thanks for clarification.

Regards

Markus Wanner



signature.asc
Description: OpenPGP digital signature
--
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from 
the latest Intel processors and coprocessors. See abstracts and register 
http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] Built-in Svn client code crashing

2013-10-04 Thread James Turner

On 4 Oct 2013, at 19:29, Markus Wanner mar...@bluegap.ch wrote:

 That smells like trouble from a packaging standpoint. It's usually not
 acceptable, because most of the time, the integrated library isn't
 getting the amount of support the original does.
 
 For Debian, I'll certainly have to consider reverting that change.

Well, that might be quick tricky - the replacement code is a tiny subset of 
what the libsubversion code does, and behaves differently -  which enables 
various features and different APIs which I'm planning to use going forwards. 

Just to be clear, I've replaced libsubversion, which is a full, read+write svn 
client library with support for history, logging, setting SVN properties and so 
on, with what is essentially a download engine which happens to speak the SVN 
protocol. (Which is the part of SVN we actually use). I haven't taken a copy of 
the libsubversion and forked/edited/trimmed it - it's completely unrelated 
code. 

Does that still cause problems under to policies described above?

Regards,
James


--
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from 
the latest Intel processors and coprocessors. See abstracts and register 
http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] Built-in Svn client code crashing

2013-10-03 Thread James Turner

On 3 Oct 2013, at 12:38, Saikrishna Arcot saiarcot...@gmail.com wrote:

 Just to check, is the built-in SVN code effectively replacing the 
 external SVN code (from libsvn-dev), or does it add something?

Replaces it - one of the big motivations is that the libsvn dependency is 
becoming increasingly complex to support. (Since libsvn depends on APR, amongst 
other things)

In Git now, all references to libsvn are gone - we always use the built-in 
code, and based on some limited feedback the crashes are gone. More testing is 
of course welcome.

Regards,
James

--
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from 
the latest Intel processors and coprocessors. See abstracts and register 
http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


[Flightgear-devel] Built-in Svn client code crashing

2013-09-20 Thread James Turner
Hello,

A few people have reported crashes when using the built-in SVN client code, 
especially on Linux (and potentially Windows too, which would be a problem, as 
we shall see). Thomas identified something strange relating to whether we were 
using built-in or the system Expat XML parser, and I finally realised the dumb 
thing I'd done, and have cleaned up the problem. Hence the refactoring that 
occurred in SimGear last night, so that we cannot (anymore) end up in a 
situation where we get the Expat headers and symbols from mismatching locations.

The guess/hope is that this was previously causing subtle memory corruption due 
to internal differences in Expat. However, it can only have been an issue on 
Linux, not Windows - since Windows will never have a system-supplied version.

So, as I've previously asked before, I really need people running from 'next' 
to try with -DSG_SVN_CLIENT=1 when configuring SimGear, move their existing 
TerraSync dir out the way, and test, test, test. I'm sure the new code isn't 
100% trouble free (in particular I think there is still the occasional time 
when it gets stuck not doing any more downloads until FG is restarted), but I 
really don't want to move forwards with the code until I have a bit more 
assurance it's not going to make everyone's setup crash 80% of the time, which 
is what some people have reported.

Note this applies even if you 'don't use' terrasync since the SVN sync engine 
is going to be used for other pieces of data as soon as it's stable. (I will be 
adding a new preference to globally control whether FG works in online/offline 
mode, of course)

Kind regards,
James
--
LIMITED TIME SALE - Full Year of Microsoft Training For Just $49.99!
1,500+ hours of tutorials including VisualStudio 2012, Windows 8, SharePoint
2013, SQL 2012, MVC 4, more. BEST VALUE: New Multi-Library Power Pack includes
Mobile, Cloud, Java, and UX Design. Lowest price ever! Ends 9/20/13. 
http://pubads.g.doubleclick.net/gampad/clk?id=58041151iu=/4140/ostg.clktrk
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel