Re: [fluid-dev] More commits (lots of windows fixes)

[josh]

Quoting David Henningsson <address@hidden>:
> address@hidden wrote:
> > Here is some status on my recent commits.
> >
> > I spent a good portion of the weekend building FluidSynth using   
> > MinGW and fixing bugs on Windows.  All MIDI drivers should support   
> > SYSEX at this point.  In addition, the TCP/IP FluidSynth shell   
> > server works now on Windows.
>
> Great, thanks for the work :-)

It was some dirty dirty Windows, but somebody had to do it!

> > One issue I have run into with the shell server though, is that it   
> > uses LF for line termination.  This seems to be contrary to the   
> > Telnet standard (should be CR LF). I'm reluctant to change it   
> > though, since
> > this could potentially break backwards compatibility with clients   
> > which use FluidSynth in this fashion.  It means that Windows   
> > telnet, and likely other telnet clients, aren't as useful as they   
> > could be.  It would be nice to just switch to using CR LF for   
> > TCP/IP shell access though.  Any opinions on that?
>
> Well, can't we just chop off any CR if we find one, at the end of each
> command? Then we would support both LF and CR LF as line terminators?

It would be nice if it was that simple.  The issue isn't with the  
client sending CRs (I added some code to handle that, basically just  
discards them), its with the server only sending LFs.  From what I  
read, the telnet protocol is supposed to be CR LF for the line  
terminators.  Unix is of course is just LF.  Not really an issue if  
someone is controlling FluidSynth from some other application via TCP,  
but if you want to telnet to that port, chances are the Telnet client  
might not work properly, unless it has support for different types of  
line endings.

If no one was relying on FluidSynth just sending LFs for its TCP/IP  
control shell, I would go ahead and switch the TCP/IP server output to  
CR LF.  Perhaps it is fine to just do so.  I doubt there are that many  
users of it and if there are, perhaps they don't depend on particular  
line endings.  Seems like telnet support would be more valuable.

> > I should probably test building FluidSynth in VC or some other non GCC
> > compiler.  I think there are probably still some issues there, but its
> > nice to have it building with MinGW!  I'll post some build directions on
> > the FluidSynth Wiki at some point soon.
>
> I remember Pedro was compiling it with VC2005/2008(?) and even had a
> pointer to where you could download that compiler for free.

Yeah, I just installed VC++ 2008 as a free download.  Lets see if I  
can get it working..

> // David


Josh