py-graph-tool: Update to version 2.57

[essandess]

Description

py-graph-tool: Update to version 2.57
* Add 311, Remove 37
* Use boost181
* Bugfix for unary_function C++17 removal

Type(s)

• bugfix
• enhancement
• security fix

Tested on

macOS 13.4.1 22F770820d arm64
Xcode 14.3.1 14E300c

Verification

Have you

• followed our Commit Message Guidelines?
• squashed and minimized your commits?
• checked that there aren't other open pull requests for the same change?
• referenced existing tickets on Trac with full URL?
• checked your Portfile with port lint --nitpick?
• tried existing tests with sudo port test?
• tried a full install with sudo port -vst install?
• tested basic functionality of all binary files?
• checked that the Portfile's most important variants haven't been broken?

[macportsbot]

Notifying maintainers:
@Veence for port cgal5.

[ryandesign]

Isn't install the default value for destroot.target?

[essandess]

It's overwritten by PG python, so this is necessary here.

[mascguy]

It's overwritten by PG python, so this is necessary here.

Well, would it be more accurate to say that this is necessary when python.pep517 is enabled? Which is the default for 3.11 and later.

I'd need to dig into the upstream project a bit, before providing any official recommendations. But if you're working around that by setting destroot.target, etc, then perhaps it might be better to simply set python.pep517 to no. (Which might allow you to eliminate at least a few overrides you're setting.)

However, I'll let our more Python-savvy folks chime in with their thoughts and recommendations.

CC: @jmroot
CC: @reneeotten

[mascguy]

I'd need to dig into the upstream project a bit, before providing any official recommendations. But if you're working around that by setting destroot.target, etc, then perhaps it might be better to simply set python.pep517 to no. (Which might allow you to eliminate at least a few overrides you're setting.)

That being said, I'm not overly concerned, so no need to hold up merging this.

[essandess]

Note: The CI and buildbots will all fail on this port for py10-graph-tool and so forth because of the way the boost ports are written: boost181, e.g., is always installed with variant +python311. There is no way to install the necessary boost variants in the Portfile.

cc: @mascguy

[mascguy]

Note: The CI and buildbots will all fail on this port for py10-graph-tool and so forth because of the way the boost ports are written: boost181, e.g., is always installed with variant +python311. There is no way to install the necessary boost variants in the Portfile.

cc: @mascguy

I'll try to take a look at a subport approach for 181 soon, as this is definitely a pain...

[essandess]

@mascguy May we please merge this PR and get a working graph-tool into the field for arm64 machines?

The boost Portfile fixes can be added later.

[mascguy]

Presently building the 3.11 subport locally on 10.15 [with trace mode], to see whether all is well. It's taking far longer than expected though.

There are also an inordinate number of warnings, enough so to utterly pollute the build log. I may suppress some of them before merging, as they make it difficult to follow the compilation itself...

[mascguy]

There are also an inordinate number of warnings, enough so to utterly pollute the build log. I may suppress some of them before merging, as they make it difficult to follow the compilation itself.

Which is probably due to the project Makefile setting -Wall and -Wextra. (With no immediately-apparent way to disable that without patching, based on a quick cursory review of the various AutoTools scripts.)

[mascguy]

p.s. You inadvertently included an update to py-absl, which you also have an open PR for. So I dropped that commit from this branch.

[mascguy]

Building a single subport takes a very long time, at least on my older MacPro. So I've only validated py311-graph-tool, but that looks good [with trace mode].

Given that there don't appear to be any dependents on this port, it might be worth dropping subports for 3.9 and prior.