At 2020-12-16T20:30:33-0600, Dave Kemper wrote: > On 12/15/20, Dorai Sitaram <ds26...@yahoo.com> wrote: > > Thanks Dave, for the suggestion. That doesn't seem to be the > > problem, however, on my machine (Ubuntu 20.10). No-argument cat > > works as expected. > > Then something peculiar to the Ubuntu groff seems to be behind this. > Ubuntu is a Debian-derived distro, and the concurrent "local timezone" > thread here points out Debian's proclivity to customize groff in > incompatible ways.
I don't think that statement is quite fair--I've known Colin for a long time and he is a diligent developer and package maintainer. Different projects face different pressures. Debian's reproducible builds project[1] is a fairly simple and crude, but nevertheless useful, form of binary verification, an assurance process for which much stronger forms are an area of active research[2]. Debian's time zone patch has confused me too, as seen in this very thread, but it's there for an understandable reason. By eliminating gratuitious differences in the artifacts generated from a source tree, we can eliminate the need for such a patch, and improve the "inspectability" of our changes to groff in day-to-day development. When I first started working on groff I used to diff build trees before and after a change with great frequency, but I lost this habit. In part that was because my confidence increased as my knowledge grew and as I wrote automated tests to reassure myself (and my fellow developers), but the most acute reason was that I simply got sick of screening out all the noise in the diff that was simply due to %%CreationDate comments and similar, and the spurious noise of changed hash key ordering, both topics covered by https://savannah.gnu.org/bugs/index.php?57218. Regards, Branden [1] https://wiki.debian.org/ReproducibleBuilds [2] https://ts.data61.csiro.au/publications/nicta_full_text/6449.pdf
signature.asc
Description: PGP signature
