Dear Massimo,
thank you for the great action.
My "requests" are the maximum and there is no need to comply to them.
About the text: which exactly happens is: "fileevents do never fire".
The event loop works, as "after 10 {set a 1};vwait a"" works.
How to detect "atfork" ?
Here is what Jan did in tcl8.5.14:
http://core.tcl.tk/tcl/info/02909e227f
Thank you,
Harald
Am 16.09.2013 23:05, schrieb Massimo Manghi:
>
> This is the messages printed by 'configure' in case Tcl version is not
> fulfilling the condition for the safe fork.
>
> ========================================================================
> WARNING!
> ========================================================================
> The Tcl notifier (and consequently the event loop) will not be working
> in Rivet running Tcl 8.5.14 and the 'prefork' MPM of Apache.
> Recommended versions are:
> - threaded builds: Tcl >= 8.5.15 (8.5 version) or Tcl >= 8.6.1
> - any non-threaded build of Tcl >= 8.5.10
> ========================================================================
>
> The message is open about the possibility the event loop is functioning,
> but definitely says with which versions of Tcl the event loop is
> expected to work.
>
> I will change it into
>
> "The Tcl notifier .... will not be working....
>
> I don't know how to detect the support of atfork at configure time.
>
> It should also be said that Rivet and Tcl are functional provided the
> programmers don't have to call the event loop. After all we all have
> been programming thousands of lines of Tcl call that work nonetheless.
>
> I should also quote the bugzilla issue that started the whole business
> about fork and the notifier
>
> There is more I'm committing: autoreconf complained about an outdated
> 'missing' file and also about the INCLUDES file that is supposed to be
> replaced by AM_CPPFLAGS
>
> why is automake using a variable hinting at C++ even though the project
> is entirely in C is beyond my comprehension.
>
>
> -- Massimo
>
> On 09/15/2013 07:13 PM, Harald Oehlmann wrote:
>> Massimo,
>> thank you for the summary and the release plans.
>>
>> I did not check the sources, if the following is already included.
>> It is a proposition. We can also say, that we do not need that.
>>
>> Inder the conditions:
>> C1) TCL threaded
>> C2) TCL Verson at least 8.5.15 or 8.6.1
>> C3) atfork not available on the system
>> Introduce a call to:
>> Tcl_InitNotifier()
>> for example in
>> Rivet_ChildInit
>> (it must be after the fork, I don't know the exact place...)
>>
>> It is also possible to introduce theis call always and not under the
>> conditions C1-C3. It does not harm. It just does not help neither:
>> C1: Unthreaded does not have the issue
>> C2: The rivet fork patch will only be in from those TCL versions
>> C3: If atfork is available, TCL does this already.
>>
>> Thank you,
>> Harald
>>
>
--
ELMICRON Dr. Harald Oehlmann GmbH
Koesener Str. 85
06618 Naumburg
Germany
Phone: +49 (0)3445 78112-0
Fax: +49 (0)3445 78112-19
www.Elmicron.de
German legal references:
Geschaeftsfuehrer: Dr. Harald Oehlmann, Jens Oehlmann
UST Nr. / VAT ID No.: DE206105272
HRB 212803 Stendal
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
