Jean-Marc Lasgouttes wrote:
> Helge> And how about files referenced from commands in ERT insets
> Helge> and/or preamble commands?
>
> AFAIK, preambles of child documents are not output, so this should not
> be a concern.
Of course. I was thinking of a macro with relative-file operation
defined
in the main document preamble, and _used_ in some document included
from a subdirectory. such as the "list a source file my way" macro.
[...]
> Well, if you do this kind of complicated things, you can probably also
> use ERT to get \subimport* ;)
If all I wanted to was to include latex - yes! Unfortunately, subimport
in ERT can't include a lyx file - I need lyx to convert the included
lyx to tex _when necessary_. I.e. when the .lyx is newer than the
existing .tex
I don't want to export all of them manually. :-(
> Seriously, I agree that this will not be supported. However, when you
> do heavy ERT, you are basically on your own. We try to have a policy
> of avoiding to add code just for the sake of making ERT work.
>
I was only hoping that the ERT that works with everything in the same
directory also could work when importing from a subdirectory.
Hmm. You claim you'll make external insets work from
subdirectories - perhaps I should try creating an external inset
for lyx documents instead?
> Helge> This is my reason for using import.sty, because it supports
> Helge> this transparently. And I guess lots of other people do
> Helge> similiar things, i.e. reference various files from ERT. And
> Helge> hope to include that lyx-file from another in another
> Helge> directory.
>
> I took a look at your patch and at import.sty, and here is my current
> thinking on it
>
> The pros:
>
> - it is a clean patch and I have no stylistic issue with it
>
> - import.sty is not too new (meaning it will be generally available)
> and is written by Donald Arseneau (this last point is important
> because it means we can rely on its quality)
>
> - it works with ERT and external insets
>
> The cons:
>
> - what your patch does not do is to make sure that, when using a temp
> dir, all files end up in the master document's
>
I know it currently only works without temp dirs - littering my
directories with all the in-between stuff latex produce.
I'm willing to fix that (and other shortcomings, if any) if that
is the key to get it into lyx. It is something I won't
bother with if I know I'll be the only user ever though.
Also note that the current patch is qt only. I have an xforms
patch for 1.2.3 that I can port easily _if_ necessary.
> - it depends on an extra package, whereas we have all the information
> to do it ourselves
Except for stuff included in ERT. :-)
> - not all packages honor \input@path (the macro that makes all that
> possible). For example skak.sty (used for the chess external
> template) does not.
>
Now that is a bad one, seems another approach might be necessary
after all. How about the following:
1. Use your way of supporting subdirectory includes, with one addition:
2. Lyx outputs a \lyxpath in the exported .tex, telling what
subdirectory
we're in. (i.e. file included from the same directory
=> \renewcommand{\lyxpath}{./} (or an empty string perhaps)
File included from subdir/ (relative to the master document)
=> \renewcommand{\lyxpath}{subdir/}
and so on.
Now anything within lyx is supported with no extra GUI
or packages, and us poor ERT-users have the hook
we need to support our own macros
properly. I could simply add \lyxpath to the filespec in my
file inclusion macros.
> - I do not think either that it will work for \bibliography{} when the
> bib file is in a subdirectory. This case can work easily (meaning I
> have to write it) with my approach
>
> - this adds yet another way of including files in the dialog. I
> suspect that this will rapidly become very confusing for users who
> do not need all these bells and whistles. I tried to find a solution
> that ``just works''.
Well, if you go fo the \lyxpath thing I have everything I need too. :-)
\lyxpath will probably not clash with other packages either.
Helge Hafting
