source-highlight: use a newer version of boost?

[Guymer (Thomas Guymer)]

"source-highlight" is failing to build because it depends on "boost171", which in turn is failing to build as reported in ticket:66686. Instead of waiting for ticket:66686 to be resolved, could "source-highlight" just be built using a newer boost version? Thanks

[reneeotten (Renee Otten)]

have you verified which newer boost port does build on your system (and is thus not affected by the ticket you mention)?

[Guymer (Thomas Guymer)]

Hi,

Thank you for getting back to me. I would say "boost176". I reinstalled all my ports yesterday and switched from Python 3.10 to Python 3.11. This meant that "port" brought in "boost176" as it is a requirement of "py311-scipy" (which is one of the ports that was built). To double check, I just ran:

user@host ~ % port installed boost\*
The following ports are currently installed:
  boost176 @1.76.0_6+no_single+no_static+python311 (active)

So yeah, let's go with "boost176" please.

Thanks,

Tom

[kencu (Ken)]

no boost versions build universal arm64/x86_64 at present.

[Guymer (Thomas Guymer)]

Then I am confused, how was I able to successfully install "boost176" yesterday but not "boost171"?

[Guymer (Thomas Guymer)]

I just ran the below commands:

sh-3.2# port install -cu boost171
--->  Computing dependencies for boost171
--->  Fetching archive for boost171
--->  Attempting to fetch boost171-1.71.0_6+no_single+no_static+python311.darwin_22.arm64.tbz2 from https://packages.macports.org/boost171
--->  Attempting to fetch boost171-1.71.0_6+no_single+no_static+python311.darwin_22.arm64.tbz2 from https://ema.uk.packages.macports.org/boost171
--->  Attempting to fetch boost171-1.71.0_6+no_single+no_static+python311.darwin_22.arm64.tbz2 from https://mse.uk.packages.macports.org/boost171
--->  Fetching distfiles for boost171
--->  Attempting to fetch boost_1_71_0.tar.bz2 from https://distfiles.macports.org/boost171
--->  Verifying checksums for boost171                                          
--->  Extracting boost171
--->  Applying patches to boost171
--->  Configuring boost171
--->  Building boost171                                  
Error: Failed to build boost171: command execution failed
Error: See /opt/local/var/macports/logs/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_devel_boost171/boost171/main.log for details.
Error: Follow https://guide.macports.org/#project.tickets if you believe there
is a bug.
Error: Processing of port boost171 failed
sh-3.2# port install -cu boost176
--->  Computing dependencies for boost176
--->  Cleaning boost176
--->  Scanning binaries for linking errors
--->  No broken files found.                             
--->  No broken ports found.
sh-3.2# port installed boost\*
The following ports are currently installed:
  boost176 @1.76.0_6+no_single+no_static+python311 (active)

The variants of both "boost171" and "boost176" are the same, yet "boost171" fails and "boost176" has been successful. I don't understand. As I see it, if "source-highlight" was to use "boost176" right now then I could install "source-highlight" successfully as I already have "boost176" installed successfully from yesterday so nothing extra would need to be built.

I apologise for my ignorance.

[kencu (Ken)]

your referenced ticket #66686 is about boost not building universal as arm64/x86_64, and no boost versions can build universal with those archs right now.

(when you built boost176, it’s not universal.)

[kencu (Ken)]

boost171 builds fine on arm macs, when last built by the buildbot at least.

​https://ports.macports.org/port/boost171/builds/

so if you’re having a problem building it, you could open a ticket for that and put up a log.

not to say that source-highlight shouldn’t use the current recommended boost… that’s a separate question, and sure, it should.

[kencu (Ken)]

one thing I do notice is that you’re using the python311 variant recently added to boost171

it’s possible that could be an issue. Ideally, these new variants are tested before being added, but ….

[kencu (Ken)]

yeah, that looks like it.

boost171 will not build with python311. The log is filled with errors like this:

:info:build ./boost/python/object/make_instance.hpp:50:31: error: expression is not assignable
:info:build             Py_SIZE(instance) = offsetof(instance_t, storage);
:info:build             ~~~~~~~~~~~~~~~~~ ^

but building boost171 with python310 is fine.

I guess nobody tested the python311 variant when it was added.

So that will be a separate ticket, I'll open that.

As boost176 is currently the default boost, then changing source-highlight to use it seems right (assuming it works, as you said it does).

[kencu (Ken)]

ticket for boost171 not building against python311 is here #67090

[Guymer (Thomas Guymer)]

Thank you very much for investigating that Ken, I had just assumed that "boost171" was not building for the same reason as the open ticket about "boost171" failing to build, rather than due to something different. I did not check to see why my build was failing to make sure that it was the same as that ticket, I apologise for that. As I see it, there are two possible ways forward to obtain "source-highlight" on a Python 3.11-only system:

1. build "source-highlight" with "boost176" instead (this ticket)
2. fix "boost171" to work with "+python311" (ticket:67090)

I prefer the first option, as it will result in fewer Boost libraries being installed on my machine, and you say that "boost176" is the default Boost currently.