[Pike 7.6.25]
Found it usefull to have the Client.pike::closed() to also look at the con->want_close flag since that flag triggers the same error as the con->closed one in decode_call().
//K
You mean in do_call? Yes, that seems correct. Are you able to check in a fix?
Martin Stjernholm, Roxen IS @ Pike developers forum wrote:
You mean in do_call? Yes, that seems correct. Are you able to check in a fix?
Oh, yes, that one. I was on the tip to ask for cvs write access, but as it is, I'm not able to check in a fix. If possible I would like to though, so maybe it would be time for me to apply for a cvs account. I've read and agree with the policies, so I'll send an e-mail to marcus or someone (if the instructions at http://pike.ida.liu.se/development/ still is up to date?).
What I would like to do is fix bugs that I find (and know/learn how to fix) and could also take on to document some parts that needs to be updated in the parts that I work with. Right now it makes heave use of the remote module. I'm also about to release a few lib modules (lua, ncurses, ncurses wrapper in pike), but those are aimed for the monger repos, unless you (pike developers) want them in the standard distro. I think that the use of monger (as it matures) should be preferred though, to keep the standard installation size down, and let users add in modules as needed. The drawback being not to have a central versioning repository for the development of the monger libs.
And for a sense of personal "get to know" each other, my real name is Andreas Stenius (stockholm).
//K
pike-devel@lists.lysator.liu.se