[erlang-questions] Re: ssl and {active, once} - bug?

[Ingela Andin]

Hi!

Actually it is the new ssl implementation that has driven the http
packet support to become a supported feature in the inet driver. I
think the reason it does proably not work with the old ssl is that it
was never a supported feature when some one still actively cared about
that implementation. We are currently working hard at making the new
ssl implementation complete enough to be able to phase out the old
one.

Regards Ingela Erlang/OTP Team, Ericsson AB

2010/4/12 Tony Rogvall <tony@REDACTED>:
> Yepp. I saw your code after I wrote my mail.
> Tried it.
> The boring things is that ssl does not handle http packets (nor http header mode) in active/once mode,
> you must use passive mode. I do not really no why. Have a look in yaws to see how it's done there.
> But if you call crypto:start add the option {ssl_imp, new} the code will work :-)
>
> /Tony
>
>
>
>
> On 11 apr 2010, at 23.36, Roberto Ostinelli wrote:
>
>> 2010/4/11 Tony Rogvall <tony@REDACTED>:
>>> Send a message to the new process after the
>>> call to controlling process, wait for that message before setting {active,once}.
>>>
>>
>> this is exactly what i do, you might have read my latest mail with a
>> full module to reproduce:
>> http://www.erlang.org/cgi-bin/ezmlm-cgi?4:mss:50624:201004:fpopocbfkpppecdembbe
>>
>> as you can see, i do set the controlling process before setting
>> {active, once}, and no race can happen there.
>>
>> ?r
>
>
> ________________________________________________________________
> erlang-questions (at) erlang.org mailing list.
> See http://www.erlang.org/faq.html
> To unsubscribe; mailto:erlang-questions-unsubscribe@REDACTED
>
>