Re: Getting link destination errors with index

[Valery Ushakov]

On Wed, Jun 27, 2018 at 11:49:35 +0100, Mark Summerfield wrote:

> On Wed, 27 Jun 2018 13:22:08 +0300
> Valery Ushakov <address@hidden> wrote:
> > On Wed, Jun 27, 2018 at 11:19:29 +0100, Mark Summerfield wrote:
> > 
> > > But when I use it twice, I get multiple errors:  
> > 
> > Can you provide a small self-contained example?
> 
> I just realised that I can.

Ok, so the problem here is that Lout (I guess, too lazy to RTFS)
generates unique "lout%d" @FilterOut file name for each instance of a
filtering symbol - @Pipe in this case.  But you only have one instance
of @Pipe, the one in the defintion of @X.  So they end up with the
same file name for each invocation of your utility.  Now, I guess,
again, is that lout reuses internal @Include implementation to get the
filtered data or something like that, so the end result is that on
each invocation of @X you get xrefs in the database that refer to the
same file name for different data.

The fix is to not use @Pipe, but to use @Filter directly.

  import @BasicSetup 
  def @X
    right @IndexData
  {
    def @Filter { "./fakelx.sh" "<" @FilterIn ">" @FilterOut }

    @F { @IndexData }
  }

  @X{ Classes/Inheritance/Mixins } # NB: not quoted because filtered

And change reallx.sh to use standard input instead of command line
arguments.

If that's too much of a trouble, you can

  import @BasicSetup 
  def @X
    left @IndexData
    right @Output
  {
    def @Filter { "./fakelx.sh" @IndexData "<" @FilterIn ">" @FilterOut }

    @F { @Output }
  }

  { "Classes/Inheritance/Mixins" } @X {} # NB: quoted

Note that you cannot use right @IndexData here (and ignore it when
passed as input) b/c you also need to refer to it inside @Filer to
pass it on the command line and that creates a loop.

I hope you can adapt this to your actual usage.


-uwe