[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [leafnode-list] 1.9.19 status



Matthias Andree wrote:
> 
> otr schrieb am Mittwoch, den 02. Mai 2001:
> 
> > > To do that, one would better encapsulate the entire I/O stuff in an own
> > > submodule that handles timeout (select or poll-based maybe) and buffering.
> >
> > In fact, there are two items in this suggestion:
> > 1/ Suppress the use of longjmp()
> > -  Very easy to do. I set "usesupplement" to -2, and I have only one test to
> > add.
> 
> Wee, I'm trying to get rid of global variables.

I am talking *only* about 1.9.xx new releases.

> > 2/ Combined which the use of "signal delivery blocking", I can also be sure
> > that the current downloaded article will not be truncated.
> 
> Signal semantics infer major portability problems (SysV vs. BSD semantics).

I now, I now... But "Signal Sets" are common to both SysV and BSD.

*But* this afternoon I realized that I don't need "signal delivery blocking"
features to do what is needed !! (my apologies ...)

One can suppress lonjmp() and associated bugs in a much simpler way  :-))
only for the 1.9.xx releases of course ;-)

My interrupt routine is as simple as this one:

--------------------------------------------
static void my_poor_work( int signo ) 
{
    if (signo == SIGQUIT) gotonextgroup = 1;
    else usesupplement = -2;
}
--------------------------------------------

> > -  As another benefit, the file var/spool/news/leaf.node/my.news.server is
> > correctly updated. (I consider this as a bug).
> 
> True.
> 
> > Well, if I have 1000 articles to download in a group, and my phone connexion
> > is slow and I am in a hurry..., I am happy to be able to interrupt this
> > group download and go to the next.
> > It takes also only about ten lines for the code.
> 
> SIGQUIT is free and can be conveyed from the keyboard (Ctrl-\).
> Will implement for 2.0. Objections?

No objections of course ! 8-)

Greetings

-- 
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx -- mailing list for leafnode
To unsubscribe, send mail with "unsubscribe" in the subject to the list