Re: threads vs. processes

joe_at_invalid.address
Date: 05/08/03 

Date: Thu, 08 May 2003 20:41:44 GMT

Valentin Nechayev <netch@segfault.kiev.ua> writes:

> >>> joe@invalid.address wrote:
>
> > >> I agree that the standard says that a process that is waiting
> > >> for anything other than a CPU is blocked. However, it looks to
> > >> me like the standard does not allow that to happen when
> > >> reading/writing a descriptor that select() has said is
> > >> active. It says that when select() says a descriptor is ready
> > >> to read, a call to an input function on it with O_NONBLOCK
> > >> clear will not block.

> >> If that call had itself been the I/O operation, it would not
> >> have blocked. It cannot, and does not, make future guarantees.

> > I'll just point out once more that the standard doesn't say
> > this. It doesn't say that the operation is allowed to block in the
> > future, it says it's not allowed to block. I don't see the
> > stipulation you're making in the standard.
>
> Standard is not idealistic concept written by old men of wisdom in
> elephant bone tower. At least for Posix and SUS, standard is written form
> of facts of a few widely used systems, facts of practice.
> Your reading is incorrect because it is incorrect for real systems.
> For real systems, the following is true:
> - If file object is declared as 'ready for reading', it is ready for
> reading without blocking until the first consequent read call is done
> and if data isn't discarded by another call on this object.

Ok, but what makes you think I disagree with that?

> - If file object is declared as 'ready for writing', it is ready for
> writing without blocking until the first consequent write call is
> done and if non-blocking writeability isn't discarded by another
> call on this object.

Same comment here.

> If you can't find this in standard, either standard is flawed or you
> make flaw. But your reading is incorrect in any case.

Well, we're wrong together then, because I agree with what you said
here. I'm not sure what you thought I was saying.

Joe

Relevant Pages

  • Re: Application Speed Problem - VERY weird
    ... MDE front ends not MDB ... If the DB is reading in a lot of records, ... The data is linked with the standard Access "link ... add that we have only ever used the database on Access 2K. ...
    (microsoft.public.access.tablesdbdesign)
  • Re: home GTT - 75 grams glucose?
    ... hospital) and taking no medicines that could affect your blood glucose. ... Those are the standard numbers I've always seen. ... a significantly higher reading than just using the 30m/1h/2h/etc. ... Good doctors KNOW they are probably going to miss the peak just ...
    (alt.support.diabetes)
  • Re: home GTT - 75 grams glucose?
    ... hospital) and taking no medicines that could affect your blood glucose. ... Those are the standard numbers I've always seen. ... a significantly higher reading than just using the 30m/1h/2h/etc. ... Good doctors KNOW they are probably going to miss the peak just ...
    (alt.support.diabetes)
  • Re: asrock, problem with nic after windows-boot - Exact Opposite issue the OP is having
    ... concerning Reading the Startup "Buffer" of the Bootup ... Time to read those startup scripts, ... are identified as STANDARD, INFORMATIONAL and HISTORIC, etc.? ...
    (comp.os.linux.networking)
  • Pam_radius readme
    ... Can anyone tell me whether I am reading this incorrectly. ... The sample config shows the username and password in plain ... Is this incorrect? ...
    (FreeBSD-Security)