Re: Patch for ice-9/format.scm (~&)

[Matthias Koeppe]

Marius Vollmer <address@hidden> writes:

> Matthias Koeppe <address@hidden> writes:
> 
> > This little patch fixes the behavior of the ~& (fresh line) format
> > directive.  FORMAT used to maintain the current column in a global
> > variable, thus formats to different ports and other means of output
> > did not mix very well (see my recent test-suite addition).  This patch
> > makes FORMAT get the initial output column from the port via
> > PORT-COLUMN. 
> 
> Thanks!  I changed your patch slightly.  I use this
> 
>   (and-let* ((col (port-column port)))        ; get current column from port
>     (set! format:output-col col))
> 
> to initialize output-col.  This makes it carry the state form one call
> to format to the next when the port does not maintain its own columns.
> What do you think?

I think that it is a bad idea.  Format should not carry global state.
It will break unexpectedly if you format to _two_ ports that don't
maintain their own columns.

Would you also commit the related test-suite additions I posted
recently?  (Sorry for the strange messages without subject etc.; I was
having trouble with the mailer.)

-- 
Matthias Köppe -- http://www.math.uni-magdeburg.de/~mkoeppe