[tex-live] documentation for "geometry"

[Gerben Wierda]

>>>>>> "Karl" == Karl Berry <karl at freefriends.org> writes:
>
>   >     In TL there are many files which cannot be accessed by
>   > texdoctk yet.  I admit I don't see that as sufficient reason to
invent strict new directory structures.
>
>   > It seems to me that what would be most general/useful is to create
doc-related data files which list what files are available, what
language they are in, and so on.  texdoc/texdoctk could be
>   > modified to read such files much more easily than we change the
whole tree around.
>
> Yes but we need some conventions for the file system to avoid name
clashes.  There are too many files "manual.pdf" yet.
>
> When I suggested a scheme like
>
>      texmf/doc/<category>/<package>/...
>
> I always thought about whether we need <category> at all.  But I think
that we need it, we can no longer rely on the fact that filenames are
unique on a system.  We have 55,312 files in TL already.
>
> XML files don't help.  We need a directory standard to avoid name
clashes.  And Context and LaTeX are different worlds.  I think that Hans
should name his files as he likes without trying to avoid name clashes
with LaTeX files.  This can only be achieved with a proper file system
standard.
>
> And if we are in luck, the number of files increases linearly with time.
Then we have 100,000 files in ten years.  But I fear that it increases
exponentially.  Sigh.
>
> We really depend on directories.

And those *are* uniquely named, generally. So why not use the directory
names as information and combine that with a set of fixed standard names
(like manual, readme, etc)? When the file is called

    doc/latex/geometry/manual.pdf

"dtexdoc geometry" could open it and "dtexdoc --all geometry" could open
all PDF files in the directory. A "dtexdoc -all latex" could open all
LaTeX docs (a lot) and a "dtexdoc --all context"  could open all the
ConTeXt docs (which have names I always forget).

Just an idea but it might be very comfortable.

G

G