[hylafax-devel] Re: Hylafax's Goals & Suitable Modems

[Dmitry Bely <dbely@mail.ru>]

"Harald Pollack" <Harald.Pollack@DATAnews.at> writes:

Hello, Harald!

> >> >-Class 1 isn't suitable for a server based fax solutions due to strict timing
> >> > requirements. This eliminates about half the modems listed.
> >> 
> >> Obviously, you have never tried it. class 1 is the one and only choice
> >> for reliability.

Theoretically yes. But practically it is much harder to write *correctly*
working Class 1 implementation than Class 2/2.0 one.

> >... if Class1 implementation in the fax software is error-free (or at least 
> >contains less errors than comparable Class 2.x firmware :-))
> 
> What do you mean with 'error-free' ? Has no bugs ? Is not in accordance to T.4/T.6/T.30 ? 

It there any difference? Software that violates standards
(T.4/T.6/T.30/...) is one huge bug :-))) 

> >I am not sure that Hylafax's Class 1 implementation complies with that criterion
> 
> Simply try it. Step by step (first without ECM :-) it should be possible to reach a stable 
> implementation. In my opinion, it is easyer to implement class 1 than class 2/2.0 ...

You want me to create new Class1 software from scratch? Oh no, it too
much. But I'll try to fix Hylafax Class1 bugs. I already switched my modem
to the Class1 mode :-)

> Main work is done by interrupt service routine of /dev/tty, you only supply correct frames 
> and receive frames which you have to interpret. Timing is less harmfull, because timeout 
> cycles and retrycounts are well documented and of such kind, that the slowest CPU you can 
> imagine, will do its job. Use a dedicated thread for handling i/o and another for 
> read/write to harddisk.

Unfourtunately Hylafax is not multi-threaded. So timing could be critical
sometimes (IMHO) ...

> >-- just remember "GSM" discussion here regarding Class 1 timing issues.
> 
> I have never understand, what the real problem was :-)

I had the strong feeling that T4 timer worked wrong. To find out is this true or 
not is in my todo list :-)

Hope to hear from you soon,
Dmitry



____________________ HylaFAX(tm) Developers Mailing List ____________________
 To unsub: mail -s unsubscribe hylafax-devel-request@hylafax.org < /dev/null