On Wed, Jan 01, 2020 at 08:36:24AM +0900, Akira Yokosawa wrote:
> On Mon, 30 Dec 2019 16:06:27 -0800, Paul E. McKenney wrote:
> > On Sun, Dec 29, 2019 at 07:21:42PM +0900, Akira Yokosawa wrote:
> >> This patch set was inspired by your change log of commit 686915e83f79
> >> ("utilities/autodate: Convert current tag to release/edition text")
> >> saying:
> >>
> >> This should really use a canned library to convert numerals to
> >> English-text ordinals, but none of the ones I found worked as advertised.
> >>
> >> My guess is the ones you found are assuming bash and don't work with
> >> dash. You know, dash doesn't handle arrays.
> >
> > It was instead that my versions of python didn't know about the
> > advertised library, possibly because I didn't know to tell them.
>
> Hm, installing python (or python3 nowadays?) packages can be tricky.
> (system/user install, installing by pip/pip3 or apt ...)
Ah, as the old saying goes, I have much to learn!
> >> Patch #1 is my answer and delegates the handling of ordinals to
> >> a LaTeX package "fmtcount".
> >
> > But I like your approach better in any case, thank you!!! ;-)
> >
> >> Patch #2 is a minor update so that a repository with untracked files
> >> is treated as clean.
> >
> > My use case for wanting the current behavior is when I add a file,
> > add references to it, but forget to do "git add". The "(m)" is
> > a hint that I forgot something.
>
> Ah, current behavior does help such a situation!
>
> > But what is the countervailing use case?
>
> Not much of a use case, but for example, random experimental outputs
> under CodeSamples/ not covered by .gitignore will generate the "(m)",
> which looks irrelevant.
I like your suggestion below of creating directories for experiemental
not-to-be-committed results using .gitignore, and would of course welcome
a patch.
> But my view point was of receiving side of your updates.
> Now I know you've relied on the current behavior and changing it
> should be regarded as a regression.
OK, we should be good as-is, then.
> > Maybe there is a way to
> > solve both problems.
>
> Let's keep the current behavior.
Sounds good!
> >> Patch #3 expands the use of tag info and displays it on every page
> >> at the bottom-right corner.
> >
> > Nice!
> >
> > I took #1 and #3, and pushed another patch that includes the most
> > recent tag for non-release builds.
>
> I thought full $description in non-release builds could be too long
> for the footer area, but the position of the watermark in the preamble
> specifies the center, and it looks reasonable. Thank you!
Very good! And thank you for coming up with the footer trick, could
save some confusion.
> > Let's talk about #2 -- perhaps
> > I should take it as is, but that would require some other warning.
> >
> > Thoughts?
>
> As mentioned above, I'm OK with keeping the current behavior.
> Adding some namespaces to .gitignore might be a better approach,
> but it is not urgent at all.
As noted earlier, this sounds good to me! One approach would be to
add the experimental directory and a README within it, with the README
describing the purpose. Then a .gitignore in that same directory could
ignore all but the README. But maybe you have a better idea.
Thanx, Paul
> Thanks, Akira
>
> >
> > Thanx, Paul
> >
> >> Thanks, Akira
> >> --
> >> Akira Yokosawa (3):
> >> utilities/autodate: Use 'fmtcount' package for ordinals
> >> utilities/autodate: Ignore untracked files in 'git status'
> >> utilities/autodate: Put commit-id in footer by 'draftwatermark'
> >>
> >> FAQ-BUILD.txt | 6 ++++--
> >> perfbook.tex | 9 +++++++++
> >> utilities/autodate.sh | 35 ++++++++++++-----------------------
> >> 3 files changed, 25 insertions(+), 25 deletions(-)
> >>
> >> --
> >> 2.17.1
> >>
>
