TeX Hour: Thu 27 May: Understand + refactor TeX Live: more flexible + agile

[George N. White III]

On Thu, 27 May 2021 at 18:37, Norbert Preining <norbert at preining.info>
wrote:

> Hi Jonathan,
>
> On Wed, 26 May 2021, Jonathan Fine wrote:
> > Don Knuth wrote TeX to be archival. To give identical results from
> > identical inputs. My TeX source file is just one of many inputs.
> Recording
> > and exactly reproducing the other inputs is vital, to fully benefit from
> > the hard work of Don Knuth that made TeX archival.
> >
> > One goal is to make it much easier to archive all the resources used
> when a
> > document is typeset. And at a different time and place, and on a
> different
>
> Unfortunately that is getting rather difficult wrt to latex or xetex etc
> - sty files can easily be found by using -recorder and copy all recorded
>   files to the cwd
> - fonts are much more tricky, as one needs to archive all the metafont
>   sources at that time, or the ttf/otf fonts used in case of xetex.
>

Or type1 fonts.   Some documents used commercial system or printer fonts
that aren't freely available.

  Which, when used with lookup by filename versus lookup by font name
>   can be tricky.
> - engines: some engines undergo changes that also might have an effect on
>   the layout, so you need to archive also the engines' source code
>   and be able to rebuild the sources
>
> All in all, a goal that is hard to achieve completely. But with archival
> of the runtime files (sty, tex, ...) one goes already a long way, and it
> is rather trivial thanks to -recorder.
>

My experience has been that many active TeX users spend their careers
using a small number of templates acquired in grad school or from
publishers and then expanded with a non-decreasing list of \includepackage
statements.  This works because they can just install TeX Live to get
all those packages along with a few they have been copying from old
system to new system.

Archiving is more likely to be useful if care is taken to choose stable
and well supported runtime files and fonts.  It would help to have
curated collections chosen to meet the needs of major publishers,
documentation systems like Rd, etc.  The collections should provide
"core" document classes to cover a majority of use cases.

>
> Fonts archiving gives you a little plus-alpha, but it is already a bit
> more involved.
>

There are some well-supported high quality free font collections that
mostly stay backwards compatible, but few users are in a position to
evaluate font quality.

>
> Engines archiving is only for the freaks.
>

I think considerable effort goes into avoiding breaking old documents
with new engine versions.   Building old engines on current toolchains
may be more difficult than updating old documents to work with current
toolchains.

It would be useful to have a few collections of old documents for
systematic testing.

-- 
George N. White III
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/tex-live/attachments/20210527/dcda9417/attachment.html>