snac @2.18: use the right compiler and flags

[ryandesign (Ryan Schmidt)]

snac isn't UsingTheRightCompiler nor flags:

​https://build.macports.org/builders/ports-10.9_x86_64-builder/builds/217714/steps/install-port/logs/stdio

cc -g -Wall -D st_mtim=st_mtimespec  -I/opt/local/include -c main.c
cc -g -Wall -D st_mtim=st_mtimespec  -I/opt/local/include -c data.c
cc -g -Wall -D st_mtim=st_mtimespec  -I/opt/local/include -c http.c
cc -g -Wall -D st_mtim=st_mtimespec  -I/opt/local/include -c httpd.c
cc -g -Wall -D st_mtim=st_mtimespec  -I/opt/local/include -c webfinger.c
cc -g -Wall -D st_mtim=st_mtimespec  -I/opt/local/include -c activitypub.c
cc -g -Wall -D st_mtim=st_mtimespec  -I/opt/local/include -c html.c

The port is manually setting use_configure no and not doing any of the manual work needed to support this properly.

The makefile portgroup could help.

[artkiver (グレェ)]

Ah, thanks. I will look into this further. I figured there was probably a better way to do this earlier, but I admit I am not super up to speed on portgroups. I will do additional research and experimentation and hopefully get this smoothed out a bit!

[artkiver (グレェ)]

So, snac 2.19 was released, which reminded me to look into this in more detail.

​https://guide.macports.org/chunked/reference.portgroup.html is a bit sparse as far as information on the makefile PortGroup.

​https://github.com/macports/macports-ports/blob/master/_resources/port1.0/group/makefile-1.0.tcl

Is I guess a useful reference insomuch as it is the TCL being invoked, though I am not sure how well I am parsing it as a human.

Regardless, doing some minor edits to the Portfile

e.g. PortGroup makefile 1.0

Still seems to install OK after I do a port -v clean.

In reading, I thought about adding:

build.args-append   CFLAGS=-g -Wall -D st_mtim=st_mtimespec

And it doesn't seem to cause harm?

I even explored removing the

patchfiles          Makefile.patch
post-patch          {reinplace "s|/usr/local|${prefix}|g" ${worksrcpath}/Makefile}

As well as attempting to add some other things to build.args-append

Such as:

PREFIX=/opt/local
and PREFIX_MAN=/opt/local/share/man

But I may have been a bit too ambitious with some of that and seemed to get some errors back when running lint --nitpick for example.

So, at the moment, I have a rather more conservative Portfile with some changes, which seems to function OK; but I am not sure if it addresses all the issues raised here.

I'm attaching it for reference though. Doubtlessly, it can be improved upon somewhat, but I figure best to update the Trac ticket for the time being as no one has replied on IRC to my inquiry yet.

[artkiver (グレェ)]

Portfile for net/snac which adds the makefile PortGroup, updates the version and checksums for 2.19, updates the long_description and tests OK locally.

[artkiver (グレェ)]

Well, what "worksforme" on my local system apparently made the build bots unhappy when I submitted a PR:

e.g.

​https://github.com/macports/macports-ports/actions/runs/4021961917/jobs/6911272754 ​https://github.com/macports/macports-ports/actions/runs/4021961917/jobs/6911272700

I can, presumably, revert to an even more conservative Portfile version bump and I am guessing it would pass the CI checks OK, but probably not adequately address the concerns raised here.

[artkiver (グレェ)]

A more conservative PR was submitted here: ​https://github.com/macports/macports-ports/pull/17799

To update snac to 2.23 to be in alignment with the upstream project.

Thankfully that PR passed the CI build bot checks at least unlike my previous documented attempt, but omits the recommendations and other efforts described in this ticket, even if it does at least address the concerns raised in this one:

​https://trac.macports.org/ticket/66989

So, for the time being I am updating this ticket and leaving it open as a reminder that I have additional effort to put in so as to refactor the Portfile to utilize the makefile PortGroup.