On Wed, 21 Oct 2026 at 18:44, Paul Eggert via tz <[email protected]> wrote:

> I'm mostly worried about dead links in our .html files.
>

Correct.  The .html files are more "live documentation" about how the
database is structured, similar or related projects, etc. so they tend to
be seen by more people with passing interest in the database than serious
researchers who have other tools at their disposal like the Wayback Machine.


> If the target page isn't archived we
> should leave the dead link alone; and arguably we should often do so
> even if the target page is archived, since we are reproducing a
> historical quote.


Yes, while the linkrot isn't ideal, some amount is unavoidable, and the
commentary in the data files is already tagged in a fairly consistent way
which provides dates that should help future researchers find any
contemporaneous mailing list discussions if necessary.

For comments in other files, in a few cases I have replaced dead links
> with links to web.archive.org. Doing that properly requires knowing
> which archive.org versions to link to; this generally can be deduced
> from the date of the comment.
>

I will add that, *if* there is any useful semi-recurring task here that one
*could* automate, it would be quite a light one and not really part of this
project, though supportive: Whenever there is a new tz release, it would be
nice to ensure that the handful of new links throughout end up being
submitted to Wayback for archival if they haven't been already.  This would
help those future researchers when they do inevitably come across a dead
link, helping stem the effects of future linkrot.

--
Tim Parenti

Reply via email to