Re: scs2ascii is broken?

On Mon, 18 Mar 2002, Scott Klement wrote:

> To be honest with you, James, I don't really understand the whole "passed
> outfile" approach.   Not that it hurts anything, but what can you achieve
> this way that you couldn't also achieve by simply by redirecting stdout to
> a file?

Yeah, I can't really think of anything other than it just feels right to
do:

FILE *outfile = stdout;
fprintf (outfile, "print statement");

so that the user can set outfile to whatever they may want (through some
mechanism).  But I agree that I can't think of anything that might want to
be done that couldn't be done with redirection.

> At any rate, everything else in scs2ascii just does a printf.  I wasn't
> trying to change the design of the software, just make it work again :)

Right, I think that using printf is consistent here so no complaints.

> TRANSPARENT are codes that OS/400 wants you to send to your printer
> without translating or interpreting them.
>
> I'm not sure that TRANSPARENT is very useful in the PS or PDF context.

Ok, you are probably right here.  To make them mean anything to postscript
or pdf we would probably have to write a PCL printer driver, and I think
that would be the wrong approach.  So for scs2ps and scs2pdf sending
TRANSPARENT codes to stderr is probably the right thing to do.  Do you
agree?

> My problem was, I was working on a business report on the iSeries,
> completely unrelated to the tn5250 project.  I had just done some fairly
> complicated printer coding, and was testing it for the first time.
> Nothing worked -- every time I sent it, it would just disappear.
>
> I _just_ couldn't figure out what was wrong with my RPG code!  No
> matter what I changed, nothing came out!  Finally, after tearing my hair
> out, I decided to try it on another printer, and quickly discovered that
> the problem was on my PC. :)  I was pretty frustrated.  Though, now -- 12
> hours later -- it seems pretty funny!

Oh boy, that's rough.  I guess you can send flames to me for breaking it!

I'm thinking I will put together a test suite to validate changes to cvs.
I have some great test files Frank Richter sent me and I have some of my
own, perhaps you (Scott) wouldn't mind sending me the raw scs output of
some of your nonconfidential reports?  Then I'll know together a shell
script that will run all three scs -> whatever programs on the test files
and make verification a little easier.  I'll make jpegs or something of
what the output should look like so that the generated output can be
compared.  Thoughts?

James Rich
james@eaerich.com