Linux man-pages PDF book

Hi Deri,

Since the conversation in groff's bug report was getting off-topic, I'm opening this email thread.

The makefile target was recently renamed from build-book to build-pdf-book (for consistency reasons).

I also reorganized the files to put them inside the build system. Here are now the files most relevant for generating the book:

	$ find \
		share/mk/configure/build-depends/texlive-fonts-extra/ \
		share/mk/build/fonts/ \
		share/mk/build/pdf/book/ \
		-type f;
	share/mk/configure/build-depends/texlive-fonts-extra/Tinos.pfb.mk
	share/mk/build/fonts/_.mk
	share/mk/build/fonts/TINOR
	share/mk/build/fonts/tinos.mk
	share/mk/build/pdf/book/_.mk
	share/mk/build/pdf/book/prepare.pl
	share/mk/build/pdf/book/front.roff
	share/mk/build/pdf/book/an.tmac

I removed the Tinos.pfa file, since now I generate it at build time from the Tinos.pfb file (installed with Debian's texlive-fonts-extra).

Either the reorganization, or the generation of the font results in a new warning:

troff:<standard input>:1649: error: cannot load font 'TINOR' to mark it as special

Since I didn't notice any changes in the PDF visually, and I removed a gigantic binary file, I proceeded with it, even with the new warning. It would be great if we can do something similar to get rid of the TINOR file, and generate it at build time.
I would also like to do the same with the Unifont font, which I haven't yet added because I hate adding gigantic binary files to the repo.

There's also the page issue. Now it seems to reset the page number for every TH.

Have a lovely day!
Alex

There's also the page issue. Now it seems to reset the page number for every TH.

Another thing I want to do, as mentioned in the groff bug, is to remove our an.tmac. Is there any change we can use a minimal definition that we can cat(1) to the system an.tmac? Or maybe instead of cat(1)ing it, tell groff(1) to use our definition in addition to -man.

Quoted 5 lines · Alejandro Colomar

Another thing I want to do, as mentioned in the groff bug, is to remove our an.tmac. Is there any change we can use a minimal definition that we can cat(1) to the system an.tmac? Or maybe instead of cat(1)ing it, tell groff(1) to use our definition in addition to -man.

Deri had been using the cover page document to append to man(7) macros definitions, which basically accomplished this.

I've since refactored everything that hyperlinked book generation needed in that respect into groff's "an.tmac" (in Git), leaving the cover page to do only cover page things.

https://git.savannah.gnu.org/cgit/groff.git/tree/doc/GMPfront.t.in

Quoted 2 lines · Alejandro Colomar

There's also the page issue. Now it seems to reset the page number for every TH.

This might be a simple issue. Make sure that you're passing groff (or troff) an option to set the C register to a true value.

groff_man(7):
-rC1 Number output pages consecutively, in strictly increasing
sequence, rather than resetting the page number to 1 (or the value of register P) with each new man document.

Regards,
Branden

Hi Branden,

Quoted 11 lines · G. Branden Robinson

Quoted 2 lines · Alejandro Colomar

There's also the page issue. Now it seems to reset the page number for every TH.

This might be a simple issue. Make sure that you're passing groff (or troff) an option to set the C register to a true value.

groff_man(7):
-rC1 Number output pages consecutively, in strictly increasing
sequence, rather than resetting the page number to 1 (or the value of register P) with each new man document.

Hmmmm. Maybe I should follow v7's tradition and restart the page number at each TH. Let's call it an accidental improvement, and not a regression. :)

Although it would be interesting to learn when/why this changed.

Quoted 5 lines · G. Branden Robinson

I've since refactored everything that hyperlinked book generation needed in that respect into groff's "an.tmac" (in Git), leaving the cover page to do only cover page things.

https://git.savannah.gnu.org/cgit/groff.git/tree/doc/GMPfront.t.in

Hmmm. I notice that your cover page has a few things that we have as part of the prepare.pl script:
<https://git.savannah.gnu.org/cgit/groff.git/tree/doc/GMPfront.t.in#n7>
<https://git.savannah.gnu.org/cgit/groff.git/tree/doc/GMPfront.t.in#n42>
<https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/tree/share/mk/build/pdf/book/prepare.pl#n86>

Maybe we could do the same, to reduce the work of prepare.pl?

Our front page is also clean from an.tmac stuff. We have the an.tmac fork here:
<https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/tree/share/mk/build/pdf/book/an.tmac>

And the front page is:
<https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/tree/share/mk/build/pdf/book/front.roff>

However, our an.tmac is not for appending, but for replacing man(7). :( I'd like to get rid of that an.tmac fork. Does your message mean that if I use groff git HEAD to build our book I can just drop the fork and use man(7), and groff(1) will do the right thing?

Also, what does .t mean (in GMPfront.t.in)? I changed the file extension to .roff (so, <front.roff>) in the Linux man-pages, as it's just a roff(7) file.

Hmmmm. Maybe I should follow v7's tradition and restart the page number at each TH. Let's call it an accidental improvement, and not a regression. :)

Although it would be interesting to learn when/why this changed.

Quoted 10 lines · Alejandro Colomar

Quoted 4 lines

groff_man(7):
-rC1 Number output pages consecutively, in strictly increasing
sequence, rather than resetting the page number to 1 (or the value of register P) with each new man document.

Hmmmm. Maybe I should follow v7's tradition and restart the page number at each TH. Let's call it an accidental improvement, and not a regression. :)

Although it would be interesting to learn when/why this changed.

It seems to have been here:

The makefile target was recently renamed from build-book to build-pdf-book (for consistency reasons).

I also reorganized the files to put them inside the build system. Here are now the files most relevant for generating the book:

	$ find \
		share/mk/configure/build-depends/texlive-fonts-extra/ \
		share/mk/build/fonts/ \
		share/mk/build/pdf/book/ \
		-type f;
	share/mk/configure/build-depends/texlive-fonts-extra/Tinos.pfb.mk
	share/mk/build/fonts/_.mk
	share/mk/build/fonts/TINOR
	share/mk/build/fonts/tinos.mk
	share/mk/build/pdf/book/_.mk
	share/mk/build/pdf/book/prepare.pl
	share/mk/build/pdf/book/front.roff
	share/mk/build/pdf/book/an.tmac

Hi Alex,

Quoted 13 lines · Alejandro Colomar

Quoted 5 lines · G. Branden Robinson

I've since refactored everything that hyperlinked book generation needed in that respect into groff's "an.tmac" (in Git), leaving the cover page to do only cover page things.

https://git.savannah.gnu.org/cgit/groff.git/tree/doc/GMPfront.t.in

Hmmm. I notice that your cover page has a few things that we have as part of the prepare.pl script:
<https://git.savannah.gnu.org/cgit/groff.git/tree/doc/GMPfront.t.in#n7>
<https://git.savannah.gnu.org/cgit/groff.git/tree/doc/GMPfront.t.in#n42>
<https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/tree/share/mk/build/pdf/book/prepare.pl#n86>

Quoted 1 line

Maybe we could do the same, to reduce the work of prepare.pl?

I didn't look closely at that complicated Perl script, but it seems likely. Essentially, anything that didn't need to be parameterized (i.e., lines you write out with Perl but don't need to do any Perl variable interpolation in), I would keep in a plain text document.

Quoted 3 lines

Our front page is also clean from an.tmac stuff. We have the an.tmac fork here:
<https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/tree/share/mk/build/pdf/book/an.tmac>

Might be worth diffing that with groff Git HEAD.

Quoted 2 lines

And the front page is:
<https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/tree/share/mk/build/pdf/book/front.roff>

Yup, that's pretty clean and focused.

Quoted 4 lines

However, our an.tmac is not for appending, but for replacing man(7). :( I'd like to get rid of that an.tmac fork. Does your message mean that if I use groff git HEAD to build our book I can just drop the fork and use man(7), and groff(1) will do the right thing?

I think so, and want to know if it doesn't. Also, fair warning, Deri said he observed a CRAZY bad performance regression when building the Linux man-pages book with groff Git HEAD. If you can reproduce that, then I have some work to do. Let me know.

Quoted 4 lines

Also, what does .t mean (in GMPfront.t.in)? I changed the file extension to .roff (so, <front.roff>) in the Linux man-pages, as it's just a roff(7) file.

It was Deri's choice. Some people have historically used the t suffix to indicate a "troff" file. I don't employ that practice, personally, because it's also popular as a suffix for "test script", and troff documents can also be rendered with nroff.

Personally, I use .roff for *roff documents I intend to be portable between AT&T/DWB troff and GNU troff, and .groff for ones that use GNU extensions.

Quoted 3 lines · Alejandro Colomar

Hmmmm. Maybe I should follow v7's tradition and restart the page number at each TH. Let's call it an accidental improvement, and not a regression. :)

I think it's a matter of taste. This issue came up last month on the groff list. As often happens with me, it turned into an episode of Unix History Detectives. :-|

https://lists.gnu.org/archive/html/groff/2024-03/msg00163.html

Quoted 1 line

Although it would be interesting to learn when/why this changed.

The default has never changed in groff as far as I know, and I'm certain I haven't personally touched it--I'd remember writing the usual 20,000 word essay with 2 dozen citations that usually accompanies my breaks with tradition.

Regards,
Branden

Quoted 5 lines

Quoted 4 lines

However, our an.tmac is not for appending, but for replacing man(7). :( I'd like to get rid of that an.tmac fork. Does your message mean that if I use groff git HEAD to build our book I can just drop the fork and use man(7), and groff(1) will do the right thing?

Quoted 4 lines

I think so, and want to know if it doesn't. Also, fair warning, Deri said he observed a CRAZY bad performance regression when building the Linux man-pages book with groff Git HEAD. If you can reproduce that, then I have some work to do. Let me know.

Can't reproduce. I've removed the an.tmac fork, and added -man to troff(1)'s invocation, and get the same times.

$ time make build-pdf-book 2>/dev/null
GROPDF		.tmp/man-pages-6.7-53-g5125d867d.pdf

real 0m13.307s
user 0m16.229s
sys 0m0.481s

$ time make build-pdf-book 2>/dev/null
GROPDF		.tmp/man-pages-6.7-53-g5125d867d-dirty.pdf

real 0m13.564s
user 0m17.060s
sys 0m0.510s

(I've built all the dependencies before hand, so that only the PDF is
built in these invocations.)
(The -dirty one uses -man.)

I forgot to mention that while I can't reproduce the performance regression, I see a regression in the navigation panel. When using groff.git HEAD's an.tmac, the navigation panel doesn't allow collapsing an entire chapter, while the Linux man-pages fork of it does allow. Both PDF books are attached to this message.

Quoted 10 lines

$ time make build-pdf-book 2>/dev/null
GROPDF		.tmp/man-pages-6.7-53-g5125d867d.pdf

real 0m13.307s
user 0m16.229s
sys 0m0.481s

$ time make build-pdf-book 2>/dev/null
GROPDF		.tmp/man-pages-6.7-53-g5125d867d-dirty.pdf

Quoted 7 lines

real 0m13.564s
user 0m17.060s
sys 0m0.510s

(I've built all the dependencies before hand, so that only the PDF is built in these invocations.)
(The -dirty one uses -man.)

Hi Alex,

Wow, I'm envious. For you, lore let those gigantic attachments through.

Quoted 5 lines · Alejandro Colomar

I forgot to mention that while I can't reproduce the performance regression, I see a regression in the navigation panel. When using groff.git HEAD's an.tmac, the navigation panel doesn't allow collapsing an entire chapter, while the Linux man-pages fork of it does allow. Both PDF books are attached to this message.

I see what you mean.

One bet is that I messed something up here and Deri's first or second hypothesis will be correct.

Here's mine.

You might need to set a register I added in this commit.

commit 5fbc22d022c07f517bc83aa9b8c97ea5536efd18

Author: G. Branden Robinson <[email protected]>
Date:   Sun Mar 3 23:57:38 2024 -0600

[man]: Add an*bookmark-base-level register.

* tmac/an.tmac: Add experimental feature to support increasing the base level of PDF bookmarks. Define register an*bookmark-base-level, initialized to zero.

(PT): Add 1 to it when producing document bookmark.

(SH): Add 2 to it when producing section heading bookmark.

(SS): Add 3 to it when producing subsection heading bookmark.

I don't know what you're using to produce the chapter headings in the navigation panel, but that likely needs to be at level 0 (or 1, depending on the convention the PDF spec uses).

I added the foregoing feature because it was obvious to me that someone might want to incorporate a man page collection into a larger work.

I admit I didn't think of the chapter organization application.

So you'd likely do something like this when building with groff Git HEAD.

'-r an*bookmark-base-level=1'

Let me know if that works out.

The foregoing is not documented because it needs field testing. Now maybe it will get some.

Quoted 14 lines

Quoted 14 lines

$ time make build-pdf-book 2>/dev/null
GROPDF		.tmp/man-pages-6.7-53-g5125d867d.pdf

real 0m13.307s
user 0m16.229s
sys 0m0.481s

$ time make build-pdf-book 2>/dev/null
GROPDF		.tmp/man-pages-6.7-53-g5125d867d-dirty.pdf

real 0m13.564s
user 0m17.060s
sys 0m0.510s

Quoted 3 lines

Quoted 3 lines

(I've built all the dependencies before hand, so that only the PDF is built in these invocations.)
(The -dirty one uses -man.)

This is very good news.

Quoted 1 line

Maybe we could do the same, to reduce the work of prepare.pl?

Hi Alex,

Feel free to have a go. These are the main differences between groff-man-pages (GM) and your book (LM):-

GM uses -mandoc not -man (by piping custom an.tmac).

LM has separate sections so .TH should use bookmark level 2 (not 1).

(git an.tmac now has:-

.\" Customize this at the command line to, for example, group multiple .\" man pages within a collection or containing document. .nr an*bookmark-base-level 0

so maybe the last stopper from using git head's an.tmac can be circumvented (except of course it won't work - overwritten by zero!!!)).

GM uses .MR to handle intra-page links, LM uses .pdfhref L. As your desperate to use Branden's code rather than mine,

I'll attach a patch which drops
an.tmac and uses the HEAD version - you may be disappointed.

I don't understand what your doing with the font, why convert to .pfa, why move it - if you have the font just put texlive path into the download file in .tmp/font/devpdf with TINOSR.

The an.patch can be applied to groff, to restore your original section hierarchy.

The shadow.patch can be applied to shadow which allows the book to be built without my custom an.tmac, using the patched version.

Quoted 2 lines

Hmmm. Your obvservation about matching the PDF navigator makes sense. Deri, was the removal of -rC1 on purpose?

I don't really know, since you removed it, so I can't answer as to your intention. The original single perl script which did the whole job had it in:-

open(BK,"| ./test-pdfmake -Tpdf -k -pet -M. -mandoc -fTINO -F/usr/share/
groff/site-font -manmark -rC1 -rCHECKSTYLE=3 > LinuxManBook.pdf");

You then deconstructed the script for your own edification, and at one stage this was the result:-

#!/bin/sh
# Copyright 2023, Alejandro Colomar <[email protected]> # SPDX-License-Identifier: GPL-3.0-or-later

(

        "$(dirname "$0")"/prepare_linux_man_book.pl "$1" \
        | preconv \
        | pic \
        | tbl \
        | eqn -Tpdf \
        | troff -Tpdf -dPDF.EXPORT=1 -dLABEL.REFS=1 -dpaper=a4 \
                -M"$(dirname "$0")" -mandoc -manmark -rC1 -rCHECKSTYLE=3 \
                2>&1 >/dev/null \
        | LC_ALL=C grep '^\. *ds ';

        "$(dirname "$0")"/prepare_linux_man_book.pl "$1";
) \
| preconv \
| pic \
| tbl \
| eqn -Tpdf \
| (
        troff -Tpdf -ms <"$(dirname "$0")"/LMBfront.ms;
        troff -Tpdf -M"$(dirname "$0")" -mandoc -manmark \
                -F"$(dirname "$0")" -dpaper=a4 ;

) \
| gropdf -F"$(dirname "$0")" -pa4;

You can see you have chosen to omit the -rC1 from the second run.

Then when
you decided a need for speed, I rewrote again a single pass version which did not rely on Branden's .MR code at all. Now you want to use similar code as used for the groff_man_pages book (back to using calls to .MR and using pdfmom --roff to satisfy forward references).

Quoted 2 lines

I removed the Tinos.pfa file, since now I generate it at build time from the Tinos.pfb file (installed with Debian's texlive-fonts-extra).

Completely unnecessary! Gropdf can use .pfb files.

Quoted 6 lines

Either the reorganization, or the generation of the font results in a new warning:

troff:<standard input>:1649: error: cannot load font 'TINOR' to mark it as special

Your _FONTSDIR must point to a directory which has a subdirectory called 'devpdf' which holds TINOR and UnifontM and a suitable 'download' file which gives the location of the actual .pf[ab] files (i.e. within texlive - no need to copy it). The difference, with/without, is in the number of glyphs not found messages.

Cheers

Deri

Quoted 8 lines

Quoted 2 lines

troff:<standard input>:1649: error: cannot load font 'TINOR' to mark it as special

Your _FONTSDIR must point to a directory which has a subdirectory called 'devpdf' which holds TINOR and UnifontM and a suitable 'download' file which gives the location of the actual .pf[ab] files (i.e. within texlive - no need to copy it). The difference, with/without, is in the number of glyphs not found messages.

I also need to understand what's that TINOR file, what type of file it is, and how I can get it.

Quoted 11 lines · Alejandro Colomar

Quoted 8 lines

Quoted 2 lines

troff:<standard input>:1649: error: cannot load font 'TINOR' to mark it as special

Your _FONTSDIR must point to a directory which has a subdirectory called 'devpdf' which holds TINOR and UnifontM and a suitable 'download' file which gives the location of the actual .pf[ab] files (i.e. within texlive - no need to copy it). The difference, with/without, is in the number of glyphs not found messages.

I also need to understand what's that TINOR file, what type of file it is, and how I can get it.

This, I can answer. It is a device-independent troff font description file. It is a plain text file in a format documented by Brian Kernighan originally in CSTR #97 in about 1982, then in the 1992 revision of CSTR #54, and of course also in the groff_font(5) man page.

The reason for the font description file is that traditionally digital fonts were (1) restrictively licensed and (2) programs that used fonts often didn't need the fonts themselves, but just descriptions of what glyphs they contained and the dimensions ("metrics") of each glyph. It is a rendering device that uses a digital font. A typesetter (or graphical display system[1]) needs only to know the metrics of the glyphs to that it can position them appropriately relative to each other.

This is why when PDF renderers substitute fonts, the result is often so terribly ugly. The glyphs of the substituted font are placed in positions appropriate to a font that may have different metrics. So they are crowded or spaced out weirdly, the kerning is off, and so on.

Quoted 4 lines · Deri

On Tuesday, 16 April 2024 02:02:28 BST Alejandro Colomar wrote:
Also I rewrote prepare.pl to not use calls to .MR, to make it faster, which is why you notice no slow down in the run since Branden released code to pdf.tmac, which affected the speed of .MR.

I want to keep an eye on this. As soon as I observe/reproduce a major performance hit (with _any_ man page collection), I mean to do something about it.

Quoted 5 lines

Quoted 3 lines

Also, how can I get
the Unifont files? I don't see any .pfa or .pfb in any unifont packages.

Do a bit of research on Peter Schaffter's "install-font.sh".

I forgot to mention afmtodit(1). The grops(1) and gropdf(1) man pages also contain step-by-step procedures

for supporting "external" fonts.

But Peter's script is vastly more convenient for pretty much everyone (except maybe distro package maintainers, who have not, as far as I know, tried to tackle the problem at all).

I don't understand what your doing with the font, why convert to .pfa, why move it - if you have the font just put texlive path into the download file in .tmp/font/devpdf with TINOSR.

I've moved it to .../devpdf and
added the download file, and it still doesn't work.

Quoted 2 lines · On Tuesday, 16 April 2024 21:57:41 BST Alejandro Colomar wrote:

I've moved it to .../devpdf and
added the download file, and it still doesn't work.

Hi Alex,

What was not clear when I wrote:-

"Your _FONTSDIR must point to a directory which has a subdirectory called 'devpdf' which holds TINOR and UnifontM and a suitable 'download' file which gives the location of the actual .pf[ab] files (i.e. within texlive - no need to copy it). The difference, with/without, is in the number of glyphs not found messages."

But instead you changed _FONTSDIR to point to devpdf.

The difference in the pages between having working and non-working fonts is seen in the attached example pages (compare with your book). These in fact are using UnifontM. TINOR provides the cyrillic glyphs and greek of the iso_8859 pages if I remember.

You can still see some gaps (which should not happen using UnifontM) they appear to be caused by an issue in one of Branden's commits, see:-

https://savannah.gnu.org/bugs/?65601

Cheers

Deri

Hi Deri,

Quoted 13 lines · Deri

Quoted 2 lines · On Tuesday, 16 April 2024 21:57:41 BST Alejandro Colomar wrote:

I've moved it to .../devpdf and
added the download file, and it still doesn't work.

Hi Alex,

What was not clear when I wrote:-

"Your _FONTSDIR must point to a directory which has a subdirectory called 'devpdf' which holds TINOR and UnifontM and a suitable 'download' file which gives the location of the actual .pf[ab] files (i.e. within texlive - no need to copy it). The difference, with/without, is in the number of glyphs not found messages."

Quoted 1 line

But instead you changed _FONTSDIR to point to devpdf.

Sorry, I misread. Fixed:
<https://www.alejandro-colomar.es/src/alx/linux/man-pages/man-pages.git/commit/?h=contrib>

Now I get some new warnings. :)

Quoted 10 lines

The difference in the pages between having working and non-working fonts is seen in the attached example pages (compare with your book). These in fact are using UnifontM. TINOR provides the cyrillic glyphs and greek of the iso_8859 pages if I remember.

You can still see some gaps (which should not happen using UnifontM) they appear to be caused by an issue in one of Branden's commits, see:-

https://savannah.gnu.org/bugs/?65601

Thanks!

Have a lovely day!
Alex

Quoted 7 lines

Cheers

Deri

Signature · 1 line

<https://www.alejandro-colomar.es/>

Quoted 1 line · Alejandro Colomar

Hi Deri,

Quoted 1 line

Now I get some new warnings. :)

(I guess that I'll need to regenerate TINOR too.)

[...]

Quoted 1 line

Quoted 1 line

Now I get some new warnings. :)

Is it a secret?

No, but they contain binary. Here's an example:

gropdf:zic.8: warning: Don't understand 'dup 0 11 -| �����B4fD�" |'

Cheers,
Alex

Quoted 3 lines · On Wednesday, 17 April 2024 11:33:58 BST Alejandro Colomar wrote:

No, but they contain binary. Here's an example:

gropdf:zic.8: warning: Don't understand 'dup 0 11 -| �����B4fD�" |'

Aah, this comes from the new font subsetting code in gropdf. I have not seen a font quite like this before, most fonts used with groff have been converted from a .ttf/.otf by running it through fontforge.If you are interested in the gory details look at the git log for the fix just committed to the groff git.

Hi Deri,

Quoted 10 lines · Deri

Quoted 3 lines · On Wednesday, 17 April 2024 11:33:58 BST Alejandro Colomar wrote:

No, but they contain binary. Here's an example:

gropdf:zic.8: warning: Don't understand 'dup 0 11 -| �����B4fD�" |'

Aah, this comes from the new font subsetting code in gropdf. I have not seen a font quite like this before, most fonts used with groff have been converted from a .ttf/.otf by running it through fontforge.If you are interested in the gory details look at the git log for the fix just committed to the groff git.

Thanks! However, I can still see that warning. I've installed groff from git HEAD:

	$ git describe
	1.23.0-1149-g5589bcd18
	$ git log -1 | head
	commit 5589bcd18776c6d236c53226a0ee30506180e6b6
	Author: Deri James <[email protected]>
	Date:   Wed Apr 17 19:25:03 2024 +0100

Improve font parsing.

The usual (for fontforge converted ttf fonts) is to use the RD, ND and NP operators within charstring definitions, however these are just named in the private subrs dictionary so could be assigned any name.

And here's the warning in the man-pages' contrib branch:

	$ make clean
	RM -rf		.tmp
	$ groff --version | head -n1
	GNU groff version 1.23.0.1149-5589
	$ make build-pdf-book -j24 >/dev/null 2>&1
	$ rm .tmp/man-pages-6.7-63-ga4f3afd28.pdf
	$ make build-pdf-book
	GROPDF		.tmp/man-pages-6.7-63-ga4f3afd28.pdf
	troff:fanotify_init.2:322: warning [page 192, 4.2i]: cannot adjust 

troff:membarrier.2:272: warning [page 475, 3.0i]: cannot adjust line statx.2:240: warning: table wider than line length minus indentation troff:syscall.2:171: warning: cannot select font 'CW' troff:syscall.2:301: warning: cannot select font 'CW'
...
gropdf:zic.8: warning: Don't understand 'dup 0 11 -| �����B4fD�" |'

Have a lovely night!
Alex

share/mk/build/pdf/book/prepare.pl +1 -1

index e23f149c6..bc922bd88 100755
@@ -1,4 +1,4 @@
-#!/usr/bin/perl -w
+#!/usr/bin/perl -wd
 #
 #	BuildLinuxMan.pl		: Build Linux manpages book
 #	Deri James (& Brian Inglis)	: 15 Dec 2022

@@ -49,16 +49,16 @@ my $dir2=$dir;
 $dir2=~tr[.][_];
 my %files;
 my %aliases;

share/mk/build/pdf/book/prepare.pl +11 -4

-	$aliases{$1}=$2;
+	$aliases{$1}="$2($3)";
 }
 
 while (my ($k,$v)=each %aliases)

@@ -68,13 +68,18 @@ while (my ($k,$v)=each %aliases)
 	}
 }
 
+while (my ($k,$v)=each %aliases)
+{
+	push(@{$revalias{$v}},$k);
+}
+
 foreach my $fn (`find "$dir"/man*/ -type f \\
 		| grep "\\.[[:digit:]]\\([[:alpha:]][[:alnum:]]*\\)\\?\\>\$";`)
 {
 	$fn=~s/\n//;
 
 	my ($nm,$sec)=GetNmSec($fn,qr/\.\d[a-z]*/);
-	$files{"${nm}.$sec"}=[$fn,(exists($aliases{"${nm}.$sec"}))?$aliases{"${nm}.$sec"}:"${nm}.$sec"];
+	$files{"${nm}.$sec"}=[$fn,(exists($aliases{"${nm}.$sec"}))?$aliases{"${nm}.$sec"}:"${nm}($sec)"];
 }
 
 my $Section='';

@@ -97,7 +102,7 @@ sub BuildPage
 	my $fn=$files{$bkmark}->[0];
 	my ($nm,$sec,$srt)=GetNmSec($bkmark,qr/\.[\da-z]+/);
 
-	my $title= "$nm\\($sec\\)";
+	my $title= "$nm($sec)";
 
 	print ".\\\" >>>>>> $nm($sec) <<<<<<\n.lf 0 $bkmark\n";
 

@@ -112,8 +117,10 @@ sub BuildPage
 		$Section=$sec;
 	}
 
-	if (exists($aliases{$bkmark})) {
+	if (exists($aliases{$bkmark}))
+	{
 		print ".eo\n.device ps:exec [/Dest /$aliases{$bkmark} /Title ($title) /Level 2 /OUT pdfmark\n.ec\n.fl\n";
+# 		print ".pdfbookmark 2 $nm($sec)";
 		return;
 	}
 

share/mk/build/pdf/book/prepare.pl +17 -21

 				my $bkmark="$1";
 				my $sec=$2;
 				my $after=$3;

@@ -145,12 +152,7 @@ sub BuildPage
 				my $dest=$bkmark;
 				$dest=~s/\\-/-/g;
 
-				if (exists($files{"${bkmark}.$sec"})) {
-					my $dest=$files{"${bkmark}.$sec"}->[1];
-					$_=".pdfhref L -D \"$dest\" -A \"$after\" -- \\fI$bkmark\\fP($sec)";
-				} else {
-					$_=".IR $bkmark ($sec)\\c\n$after";
-				}
+				$_=".MR \"$bkmark\" $sec $after";
 			}
 
 			s/^\.BI \\fB/.BI /;

@@ -175,16 +177,20 @@ sub BuildPage
 				s/\n\n/\n/g;
 			}
 
-			s/\\&\././ if m/^.TH /;
-
-			if (m/^\.TH\s+"?([-\w\\.]+)"?\s+"?(\w+)"?/) {
-
-				print "$_\n";
-
-				# Add a level two bookmark. We don't set it in the TH macro since the name passed
-				# may be different from the filename, i.e. file = unimplemented.2, TH = UNIMPLEMENTED 2
-
-				print ".pdfbookmark -T $bkmark 2 $nm($sec)\n";
+# 			s/\\&\././ if m/^.TH /;
+#
+ 			if (m/^\.TH\s+"?([-\w\\.]+)"?\s+"?(\w+)"?(.*)/)
+ 			{
+				print ".TH \"$nm\" \"$2\" $3\n";
+
+				if (exists($revalias{"$nm($sec)"}))
+				{
+					foreach my $dest (@{$revalias{"$nm($sec)"}})
+					{
+						my ($nm,$sec,$srt)=GetNmSec($dest,qr/\.[\da-z]+/);
+						print ".pdfhref M -D $nm($sec)\n";
+					}
+				}
 
 				next;
 			}

@@ -199,11 +205,8 @@ sub doMR
 	my $nm=shift;
 	my $sec=shift;
 
-	if (exists($files{"${nm}.$sec"})) {
-		return("\n.pdfhref L -D \"$files{\"${nm}.$sec\"}->[1]\" -A \"\\c\" -- \\fI$nm\\fP($sec)\n");
-	} else {
-		return("\\fI$nm\\fP($sec)");
-	}
+	return "\n.MR $nm $sec";
+
 }
 
 sub GetNmSec

tmac/an.tmac +3 -2

index d9e8232f8..46934fd62 100644
@@ -212,7 +212,7 @@
 .
 .\" Customize this at the command line to, for example, group multiple
 .\" man pages within a collection or containing document.
-.nr an*bookmark-base-level 0
+.nr an*bookmark-base-level \n[an*bookmark-base-level]+0
 .
 .\" Write a bookmark/anchor/link target $2 at hierarchical depth $1.
 .de an*bookmark

@@ -230,7 +230,8 @@
 .ds an*bookmark*utf8 \" empty
 .
 .de an*bookmark*pdf
-.  ie (\\$1 = 1) \
+.nr an:lev \\n[an*bookmark-base-level]+1
+.  ie (\\$1 = \\n[an:lev]) \
 .    pdfbookmark -T "\\$2" \\$1 \\$2
 .  el \
 .    pdfbookmark \\$1 \\$2

share/mk/build/pdf/book/_.mk +2 -2

index 3395e67e9..9f1cdc520 100644
@@ -33,12 +33,12 @@ _PDF_BOOK := $(_PDFDIR)/$(PDF_BOOK)
 $(_PDF_BOOK): $(_MANPAGES) $(_TINOS) $(MKBOOK) $(MK) | $$(@D)/
 	$(info	$(INFO_)GROPDF		$@)
 	$(MKBOOKDIR)/prepare.pl $(_MANDIR) \
-	| $(CAT) $(MKBOOKDIR)/front.roff $(MKBOOKDIR)/an.tmac /dev/stdin \
+	| $(CAT) $(MKBOOKDIR)/front.roff /dev/stdin \
 	| $(PRECONV) \
 	| $(PIC) \
 	| $(TBL) \
 	| $(EQN) -Tpdf \
-	| $(TROFF) -Tpdf -F$(_FONTSDIR) -dpaper=a4 $(TROFFFLAGS) \
+	| pdfmom --roff -F$(_FONTSDIR) -man -r an*bookmark-base-level=1 -dpaper=a4 $(TROFFFLAGS) -Z \
 	| $(GROPDF) -F$(_FONTSDIR) -pa4 $(GROPDFFLAGS) \
 	| $(SPONGE) $@
 

share/mk/build/pdf/book/prepare.pl +32 -29

index e23f149c6..bc922bd88 100755
@@ -1,4 +1,4 @@
-#!/usr/bin/perl -w
+#!/usr/bin/perl -wd
 #
 #	BuildLinuxMan.pl		: Build Linux manpages book
 #	Deri James (& Brian Inglis)	: 15 Dec 2022

@@ -49,16 +49,16 @@ my $dir2=$dir;
 $dir2=~tr[.][_];
 my %files;
 my %aliases;
-my %target;
+my %revalias;
 
 foreach my $al (`find "$dir"/man*/ -type f \\
 		| grep "\\.[[:digit:]]\\([[:alpha:]][[:alnum:]]*\\)\\?\\>\$" \\
 		| xargs grep '^\\.so' /dev/null;`)
 {
 	#$al=~tr[.][_];
-	$al=~m/^$dir\/man\d[a-z]*\/(.*):\.\s*so\s*man\d[a-z]*\/(.*)/o;
+	$al=~m/^$dir\/man\d[a-z]*\/(.*):\.\s*so\s*man\d[a-z]*\/(.*?)\.(.*)/o;
 
-	$aliases{$1}=$2;
+	$aliases{$1}="$2($3)";
 }
 
 while (my ($k,$v)=each %aliases)

@@ -68,13 +68,18 @@ while (my ($k,$v)=each %aliases)
 	}
 }
 
+while (my ($k,$v)=each %aliases)
+{
+	push(@{$revalias{$v}},$k);
+}
+
 foreach my $fn (`find "$dir"/man*/ -type f \\
 		| grep "\\.[[:digit:]]\\([[:alpha:]][[:alnum:]]*\\)\\?\\>\$";`)
 {
 	$fn=~s/\n//;
 
 	my ($nm,$sec)=GetNmSec($fn,qr/\.\d[a-z]*/);
-	$files{"${nm}.$sec"}=[$fn,(exists($aliases{"${nm}.$sec"}))?$aliases{"${nm}.$sec"}:"${nm}.$sec"];
+	$files{"${nm}.$sec"}=[$fn,(exists($aliases{"${nm}.$sec"}))?$aliases{"${nm}.$sec"}:"${nm}($sec)"];
 }
 
 my $Section='';

@@ -97,7 +102,7 @@ sub BuildPage
 	my $fn=$files{$bkmark}->[0];
 	my ($nm,$sec,$srt)=GetNmSec($bkmark,qr/\.[\da-z]+/);
 
-	my $title= "$nm\\($sec\\)";
+	my $title= "$nm($sec)";
 
 	print ".\\\" >>>>>> $nm($sec) <<<<<<\n.lf 0 $bkmark\n";
 

@@ -112,8 +117,10 @@ sub BuildPage
 		$Section=$sec;
 	}
 
-	if (exists($aliases{$bkmark})) {
+	if (exists($aliases{$bkmark}))
+	{
 		print ".eo\n.device ps:exec [/Dest /$aliases{$bkmark} /Title ($title) /Level 2 /OUT pdfmark\n.ec\n.fl\n";
+# 		print ".pdfbookmark 2 $nm($sec)";
 		return;
 	}
 

@@ -137,7 +144,7 @@ sub BuildPage
 
 			s/\\-/-/g if /^\.[BM]R\s+/;
 
-			if (m/^\.BR\s+([-\w\\.]+)\s+\((.+?)\)(.*)/ or m/^\.MR\s+([-\w\\.]+)\s+(\w+)\s+(.*)/ or m/^\\fB([-\w\\.]+)\\fR\((.+?)\)(.*)$/) {
+			if (m/^\.BR\s+([-\w\\.]+)\s+\(([\d\w]+?)\)(.*)/ or m/^\.MR\s+([-\w\\.]+)\s+(\w+)\s+(.*)/ or m/^\\fB([-\w\\.]+)\\fR\((.+?)\)(.*)$/) {
 				my $bkmark="$1";
 				my $sec=$2;
 				my $after=$3;

@@ -145,12 +152,7 @@ sub BuildPage
 				my $dest=$bkmark;
 				$dest=~s/\\-/-/g;
 
-				if (exists($files{"${bkmark}.$sec"})) {
-					my $dest=$files{"${bkmark}.$sec"}->[1];
-					$_=".pdfhref L -D \"$dest\" -A \"$after\" -- \\fI$bkmark\\fP($sec)";
-				} else {
-					$_=".IR $bkmark ($sec)\\c\n$after";
-				}
+				$_=".MR \"$bkmark\" $sec $after";
 			}
 
 			s/^\.BI \\fB/.BI /;

@@ -175,16 +177,20 @@ sub BuildPage
 				s/\n\n/\n/g;
 			}
 
-			s/\\&\././ if m/^.TH /;
-
-			if (m/^\.TH\s+"?([-\w\\.]+)"?\s+"?(\w+)"?/) {
-
-				print "$_\n";
-
-				# Add a level two bookmark. We don't set it in the TH macro since the name passed
-				# may be different from the filename, i.e. file = unimplemented.2, TH = UNIMPLEMENTED 2
-
-				print ".pdfbookmark -T $bkmark 2 $nm($sec)\n";
+# 			s/\\&\././ if m/^.TH /;
+#
+ 			if (m/^\.TH\s+"?([-\w\\.]+)"?\s+"?(\w+)"?(.*)/)
+ 			{
+				print ".TH \"$nm\" \"$2\" $3\n";
+
+				if (exists($revalias{"$nm($sec)"}))
+				{
+					foreach my $dest (@{$revalias{"$nm($sec)"}})
+					{
+						my ($nm,$sec,$srt)=GetNmSec($dest,qr/\.[\da-z]+/);
+						print ".pdfhref M -D $nm($sec)\n";
+					}
+				}
 
 				next;
 			}

@@ -199,11 +205,8 @@ sub doMR
 	my $nm=shift;
 	my $sec=shift;
 
-	if (exists($files{"${nm}.$sec"})) {
-		return("\n.pdfhref L -D \"$files{\"${nm}.$sec\"}->[1]\" -A \"\\c\" -- \\fI$nm\\fP($sec)\n");
-	} else {
-		return("\\fI$nm\\fP($sec)");
-	}
+	return "\n.MR $nm $sec";
+
 }
 
 sub GetNmSec

Hi Alex,

While testing I used a document just using the debian TinosR font, no problem. Issue occurs when both types of font are used in one document. I have now pushed the fix to groff. Please try again.

Quoted 1 line · Deri

Hi Alex,

Hi Deri,

Quoted 3 lines

While testing I used a document just using the debian TinosR font, no problem. Issue occurs when both types of font are used in one document. I have now pushed the fix to groff. Please try again.

Thanks! It works now. How can I use Debian's TinosR font? Where is it?

Have a lovely night!
Alex