Re: windres regression with control strings spanning multiple lines

[Brian Dessent <brian at dessent dot net>]

Kai Tietz wrote:

> > write such a tool just for a binutils test though.  I don't suppose such
> 
> > a tool has already been written has it ?
> 
> Couldn't we simply use windres output to .rc for this? It dumps content in
> an easy textual way.

That would be one route, to make sure that the file round-trips
rc->coff->rc and rc->res->rc with a resulting output that resembles the
input.  But you'd still have to do some filtering and processing since
on the way back you get all the default field values, e.g. for the
testcase above:

/* COFF information not part of RC
   Time stamp: 1205756079

   Type: dialog
   COFF information not part of RC
   Time stamp: 1205756079

   Name: 0
   COFF information not part of RC
   Time stamp: 1205756079.  */
LANGUAGE 9, 1

0 DIALOG 0, 0, 0, 0
STYLE 0x80880000
BEGIN
  LTEXT "This is a continued line", 0, 0, 0, 0, 0, 0x50020000
END

>From a testing standpoint I suppose it's always good to ensure your
tools are self-consistent, but something does feel a bit off in
depending on the tool being tested to verify itself.  Still, that would
be significantly better than nothing.

Brian