here is a second iteration of the ABRT's retrace server design. Some
parts are still lacking detail and cultivation, as I'm not sure if it's
worth to spend more time improving it. I need feedback.
It would help a lot if you could read it and point out suboptimal ideas
etc. New thoughts are very welcome.
We should re-send it to fedora-devel mailing list asking for general
acceptance of the idea within a few days if no disastrous response
CCing the ML.
On 10/11/2010 03:49 AM, Andrew Hecox wrote:
> - have we talked to the gcc/gdb folks to see about the technical
> feasibility of including address->symbol translations without resorting
> to debuginfo packages?
I have never been thinking about this. Do you want to reconsider
advantages/disadvantages of having "debuginfo" directly in binaries vs.
in separate packages? Or just some part of it?
> - why the async interface to the http server? I assume it's because
> you're concerned about possible response time and timeouts in web
> proxies, but just want to double-check.
> - on that note, I'd definitely encourage a synchronous interface,
> possibly ahead of an async one
I added support for synchronous interface to the design document.