1583192 - gfx/sfntly/cpp/src/sfntly/table/core/cmap_table.cc:442:21: error: reinterpret_cast from 'nullptr_t' to 'sfntly::ReadableFontData *' is not allowed

Status: RESOLVED FIXED

(Core :: Graphics: Text, defect)

Description

[Jan Beich]

FreeBSD and OpenBSD define NULL as nullptr for C++11 or later.
https://github.com/freebsd/freebsd/commit/c8ed04c26b6758354853a6bed4629f71d0d01a7d
https://github.com/openbsd/src/commit/6ecde746dea9a5d17abf3bafa06c232b9189b33b

$ c++ --version
FreeBSD clang version 9.0.0 (tags/RELEASE_900/final 372316) (based on LLVM 9.0.0)
Target: x86_64-unknown-freebsd13.0
Thread model: posix
InstalledDir: /usr/bin

$ ./mach bootstrap
$ ./mach build
[...]
In file included from Unified_cpp_gfx_sfntly_cpp_src2.cpp:20:
gfx/sfntly/cpp/src/sfntly/table/core/cmap_table.cc:442:21: error: reinterpret_cast from 'nullptr_t' to 'sfntly::ReadableFontData *' is not allowed
    : CMap::Builder(reinterpret_cast<ReadableFontData*>(NULL),
                    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
gfx/sfntly/cpp/src/sfntly/table/core/cmap_table.cc:566:34: error: reinterpret_cast from 'nullptr_t' to 'sfntly::WritableFontData *' is not allowed
                               : reinterpret_cast<WritableFontData*>(NULL),
                                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
gfx/sfntly/cpp/src/sfntly/table/core/cmap_table.cc:577:34: error: reinterpret_cast from 'nullptr_t' to 'sfntly::ReadableFontData *' is not allowed
                               : reinterpret_cast<ReadableFontData*>(NULL),
                                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
gfx/sfntly/cpp/src/sfntly/table/core/cmap_table.cc:961:21: error: reinterpret_cast from 'nullptr_t' to 'sfntly::ReadableFontData *' is not allowed
    : CMap::Builder(reinterpret_cast<ReadableFontData*>(NULL),
                    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
gfx/sfntly/cpp/src/sfntly/table/core/cmap_table.cc:969:21: error: reinterpret_cast from 'nullptr_t' to 'sfntly::ReadableFontData *' is not allowed
    : CMap::Builder(reinterpret_cast<ReadableFontData*>(NULL),
                    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
5 errors generated.

Comment 1

[Nathan Froyd [:froydnj]]

I'd recommend fixing this upstream (https://github.com/googlefonts/sfntly/) and then one can submit a pull-from-upstream patch for this bug.

Comment 2

[Jan Beich]

See bug 1330495 comment 2.

Comment 3

[Jan Beich]

Attachment: v1

Sorry, I can't submit on Phabricator due to bug 1536716.
https://treeherder.mozilla.org/#/jobs?repo=try&revision=499f2bb97e191dbd5966bc3ecbc199b09daca919

Patch generated by:

$ rg -l 'reinterpret_cast<.*>\(NULL\)' |
   xargs sed -i '' 's/reinterpret_cast<\(.*\)>(NULL)/static_cast<\1>(NULL)/'

Comment 4

[BugBot [:suhaib / :marco/ :calixte]]

:jbeich, since this bug is a regression, could you fill (if possible) the regressed_by field?
For more information, please visit auto_nag documentation.

Comment 5

[Jan Beich]

See Whiteboard field.
EDIT: Adding bug 1581145 to "Regressed by" didn't work when I've filed this bug but when I've tried again while writing this comment it did. ;)

Comment 6

[Pascal Chevrel:pascalc (PTO until April 26)]

Bug 1581145 was uplifted to 70beta and 68esr, updating tracking flags.

Comment 7

[Liz Henry (:lizzard) (relman/hg->git project)]

What needs to happen next, here? Jan do you need someone else or can you land this?

Comment 8

[Jan Beich]

I need a reviewer. Can you help?

(In reply to Nathan Froyd [:froydnj] [away until 28 October] from comment #1)

https://github.com/googlefonts/sfntly/

Looks semi-abandoned. README.md says: "This project is not developed any further. Only Bug fixes will be merged, but new features won't."
Upstream is unlikely to fix FreeBSD support on their own or switch to C++11 then convert NULL to nullptr.

Comment 9

[Jonathan Kew [:jfkthame]]

I've tried submitting a PR upstream; let's see if that gets any response.
https://github.com/googlefonts/sfntly/pull/111

Leaving ni? flag to remind me to check on this later.

Comment 10

[Val Packett]

The readme mentions a maintained fork at https://github.com/rillig/sfntly — maybe worth switching to it?

Comment 11

[Jan Beich]

This bug is making bisecting on FreeBSD harder. Maybe Skia PDF should be disabled on GTK platforms due to bug 1322653.

(In reply to greg v [:myfreeweb] from comment #10)

https://github.com/rillig/sfntly

Doesn't look actively maintained either: last change on 2018-11-13, no explicit -std=, still uses reinterpret_cast<...>(NULL). Firefox effort to use Skia PDF for printing appears stalled. Another SkPDF consumer, Project Mortar (port PDFium from Chromium), was canned in bug 1503537.

Comment 12

[Landry Breuil (:gaston)]

Fwiw, this is also harming development on OpenBSD, as m-c builds are broken since a while. Can we get the patch somehow commited to ease our lives ? Upstream seems dead/not responding to the PR..

Comment 13

[Landry Breuil (:gaston)]

Attachment: Bug 1583192 - Re-apply bug 1330495 after bug 1581145. r?froydnj

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/e847a6cb715f
Re-apply bug 1330495 after bug 1581145. r=froydnj

Comment 15

[Andrei Ciure[:aciure]]

https://hg.mozilla.org/mozilla-central/rev/e847a6cb715f

Comment 16

[Ryan VanderMeulen [:RyanVM]]

Do we need to uplift this to Beta & ESR68? If so, please nominate :)

Comment 17

[Pascal Chevrel:pascalc (PTO until April 26)]

We have shipped our last beta for 71, marking as wontfix for 71.

Comment 18

[Jan Beich]

sfntly is only built by default on Nightly and cannot be used on Linux/BSD/Solaris due to bug 1322653. Backporting isn't necessary.
https://searchfox.org/mozilla-central/rev/581466eef926/toolkit/moz.configure#842-843