Attached is the patch. It's basically just reseting auth_cred in the constructor (def __init__) of "class AccountConfig:". A one line patch. On Thu, 11 Mar 2010 01:17:08 +0100 Jens Jorgensen <jbj1 at ultraemail.net> wrote: >Perhaps you could attach a patch so all would benefit? > >pjadx at hushmail.com wrote: >> Hi, >> >> It's me responding to my message. >> I finally found out what the problem was, it's located within >> "pjsua.py" in "class Account:". >> There are here a lot of static class object, and one of them is >> "auth_cred = []", which means everytime you instanciate a new >> Account() object it has the whole credentials of previous >> instanciated Account() object. This will later create problem as > >> the C API don't support as much account as the auth_cred array >has. >> >> By just reseting auth_cred to [] in the constructor, i get rid >of >> this problem. >> Each Account() has now only ONE cred info. >> So that increasing PJSUA_MAX_ACC without increasing >> PJSUA_ACC_MAX_PROXIES is enough. >> I don't know if i did it the proper way, but well, at least it's > >> working for me. >> >> Regards, >> >> Nicolas >> >> On Tue, 09 Mar 2010 16:37:54 +0100 pjadx at hushmail.com wrote: >> >>> Hi, >>> >>> I am using the PJSIP python binding to register a *lot* of >>> account. >>> "a lot" means approximately 10 K users. >>> So, first i had to raise the PJSUA_MAX_ACC in config_site.h, >but >>> then i discovered only the first 8 account registered correctly >: >>> Above 8 the digest.username is the 8th, aka the >>> PJSUA_ACC_MAX_PROXIES'th (it means that the cred[8] will be use >as >>> >>> username for all the others 10k account minus 8). >>> >>> The problem is the following, pjsua_var.acc contains an array >of >>> PJSUA_MAX_ACC, and each one of this struct pjsua_acc contains >an >>> array of PJSUA_ACC_MAX_PROXIES credentials entry. Increasing >both >>> to 10k will make gcc refuse to compile. >>> >>> For some reason there is something here i do not understand ; >it >>> seems that each account contain the whole credentials of all >>> others >>> accounts. >>> >>> Anyway, if I want to have 10k account registering with the 10k >>> uniq >>> users, i would have to raise both PJSUA_MAX_ACC and >>> PJSUA_ACC_MAX_PROXIES to 10k. >>> But gcc refuse to compile because the structure become too big. >>> >>> I am looking for a way to make pjsip work with lot of accounts. > >>> Even if i have to modify the code source and do changement in >the >>> C >>> code (i am fluent with C). >>> I am looking at the source code since yesterday and finnally >>> decide >>> that you are the best to advise me how to solve it the clean >way >>> without too much ugly hacking. >>> >>> Any advise are welcome to resolve this situation. >>> >>> Thanks. And thanks for the good work done on PJ. >>> >>> Regards, >>> >>> Nicolas >>> >>> >>> _______________________________________________ >>> Visit our blog: http://blog.pjsip.org >>> >>> pjsip mailing list >>> pjsip at lists.pjsip.org >>> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org >>> >> >> >> _______________________________________________ >> Visit our blog: http://blog.pjsip.org >> >> pjsip mailing list >> pjsip at lists.pjsip.org >> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org >> > > >-- >Jens B. Jorgensen >jbj1 at ultraemail.net > > >_______________________________________________ >Visit our blog: http://blog.pjsip.org > >pjsip mailing list >pjsip at lists.pjsip.org >http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org -------------- next part -------------- A non-text attachment was scrubbed... Name: pjsua.py.patch Type: text/x-diff Size: 37 bytes Desc: not available URL: <http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/attachments/20100312/e58718d8/attachment.patch>
