Re: git: 28472e9722d8 - main - Revert "x11-fonts/noto*: Restructure and update"

On Mon, Mar 25, 2024 at 3:27 AM Tijl Coosemans <tijl@freebsd.org> wrote:

> On Sun, 24 Mar 2024 00:12:07 +0800 Po-Chuan Hsieh wrote:
> > On Sat, Mar 23, 2024 at 11:38 PM Tobias C. Berner <tcberner@gmail.com>
> > wrote:
> >> Am Samstag, 23. März 2024, 15:14:01 CET schrieb Gleb Popov:
> >>> On Sat, Mar 23, 2024 at 5:04 PM Po-Chuan Hsieh <sunpoet@freebsd.org>
> >>> wrote:
> >>>> The branch main has been updated by sunpoet:
> >>>>
> >>>> URL:
> >>>>
> https://cgit.FreeBSD.org/ports/commit/?id=28472e9722d808739a4d690f6d803a2b1b8f5f98
> >>>>
> >>>> commit 28472e9722d808739a4d690f6d803a2b1b8f5f98
> >>>> Author:     Po-Chuan Hsieh <sunpoet@FreeBSD.org>
> >>>> AuthorDate: 2024-03-23 14:01:07 +0000
> >>>> Commit:     Po-Chuan Hsieh <sunpoet@FreeBSD.org>
> >>>> CommitDate: 2024-03-23 14:01:07 +0000
> >>>>
> >>>>     Revert "x11-fonts/noto*: Restructure and update"
> >>>>
> >>>>     This reverts commit 3b128ff9f74b28ada2f475f0b816480b1550e5b8.
> >>>
> >>> It is good manners to explain the reason for reverting when doing it.
> >>> Even better to first talk to the original committer and ask him to the
> >>> revert.
> >>
> >> Moin moin
> >>
> >> I completely agree. Given that there has been a phabricator review open
> >> for two months, with you, sunpoet@ as reviewer, this is looking like
> poor
> >> judgement on your side.
> >>
> >> As arrowd@ points out, you should have commented on the reverted commit
> >> first, and elaborate your reasoning for a revert.
> >>
> >> Simply reverting the work of others may lead to fewer contributions.
> >> Please take this into consideration in the future.
> >>
> >>
> >> mfg Tobias
> >
> >
> > Hello,
> >
> > I'm sorry that I did not put enough information in this reverting commit
> > because I need it to be done ASAP so that less users are affected.
> > There are several reasons I reverted it by myself instead of
> > requesting tijl@ to do so.
> > First, I do not use phab review therefore I did not really review the
> patch
> > which means it is committed without maintainer approval.
> > Second, I already have my own work for noto updates which is totally in
> > different direction with tijl@'s work. I cannot simply rebase my work
> onto
> > it.
> > Third, The problem mentioned in the PR is also handled in my own work.
> >
> > Regards,
> > sunpoet
>
> I don't mind getting reverted.  I do think Noto is large enough that
> some thought should go into how we handle it in the ports tree and that
> you should have put your alternative up for discussion before committing
> it.  Here are some (possible) issues with it that should probably be
> fixed before the next quarterly:
>
> - You are using upstream distfiles that are quite large.  The distfiles
>   for the common (Latin, Greek, Cyrillic) fonts Sans, Sans Mono, and
>   Serif add up to 234MB.  These fonts will be installed by almost every
>   user that uses FreeBSD as a desktop now that DejaVu appears to have
>   been abandoned.  My proposal was to create our own distfiles and I
>   integrated that into makesum so the only extra step compared to a
>   normal version update was to upload the generated distfile to
>   freefall.  I was willing to take over maintainership in case you
>   didn't want to deal with that.  These distfiles add up to 27MB.  And
>   if we put the extra weights (ExtraBold,...) and widths (Condensed,...)
>   in a separate port like you did it would be only 2MB.
>

The committed one is my second version of noto fonts update.
The previous version is to use the individual file from github.com just
like those Noto CJK ports.
I changed to upstream tarball because it's clearer/simpler and easier to
track.
I could change it back to the previous version for a smaller distfile
footprint.
Give me several days to clean up the Makefiles.


> - Some fonts have been split into x11-fonts/<font>-basic providing Bold,
>   BoldItalic, Italic, and Regular, and x11-fonts/<font>-extra providing
>   additional styles.  Only some of these have x11-fonts/<font> as a
>   metaport currently.  This needs to be made consistent.  I think it
>   would be best to drop the metaports and move x11-fonts/<font>-basic to
>   x11-fonts/<font>.  That way pkg install <font> installs the common
>   styles for all fonts.  I think that this would be the most
>   intuitive/convenient.
>   (I was hoping to use subpackages for this instead of extra ports.)
>

The -basic/-extra structure was kept for minimal change.
I agree with you that it would be better to be consistent with all other
noto ports.
I could merge the following ports:
- noto-sans-basic/-extra
- noto-sans-mono-basic/-extra
- noto-sans-symbols-basic/-extra
- noto-serif-basic/-extra
However, what should be noto-basic's dependencies after the merge?


> - x11/noto-emoji now installs the COLRv1 version of Noto Color Emoji but
>   this isn't supported yet by graphics/cairo so emoji are broken now in
>   gtk applications.  I think Ghostscript doesn't support it either.  It
>   has to be removed again for now.
>

It is handled in PR 278019 and
ports 2878ca490586207c3cd7bb3ff94d1274f52013b0.


>
> - Besides Noto Color Emoji there's also the monochrome Noto Emoji that
>   can be useful in print.  I think we should let x11-fonts/noto-emoji
>   install the monochrome font and move the existing x11-fonts/noto-emoji
>   to x11-fonts/noto-color-emoji so the names match.
>

How about putting all emoji in the existing port (x11-fonts/noto-emoji)?