UPDATE: gpsd-3.15

[fhgwright (Fred Wright)]

I've updated this port for version 3.15, including fixes for OSX/MacPorts and also generic fixes which I've also reported upstream. I've tested the build on a MacPro running 10.9.5, a PowerBook G4 running 10.5.8, and a VM running 10.11.2. Also, all regression tests (not run as part of the normal build process) pass in all three cases (with the leap-second tweak needed to match the test data).

The attached tarball is the complete contents of .../net/gpsd/, including the two patch files unchanged from 3.14_4.

Although it looks like a 3.16 release is imminent, I decided to go ahead and send this anyway, in case there's a delay in the 3.16 release. I'll file a new ticket for 3.16 once that's ready.

[fhgwright (Fred Wright)]

Complete tarball of gpsd-3.15_0

[mf2k (Frank Schima)]

In the future, please Cc the port maintainers (port info --maintainers gpsd), if any.

​Per the guidelines, please instead attach a unified diff of the Portfile so we can easily see what changes you are proposing.

[fhgwright (Fred Wright)]

Replying to mf2k@…:

In the future, please Cc the port maintainers (port info --maintainers gpsd), if any.

Sorry, I didn't bother when I saw openmaintainer.

​Per the guidelines, please instead attach a unified diff of the Portfile so we can easily see what changes you are proposing.

Done.

I forgot to mention in the original post that this also fixes the qt and dbus variants (or more precisely, the omission thereof) to work correctly, and adds the missing Qt dependency when needed. However, for the dbus variant, I only fixed the handling of the dbus_export option, leaving the basic dbus option alone.

[fhgwright (Fred Wright)]

Since gpsd-3.16 was just released on Friday, it no longer makes much sense to bother with a 3.15 port. I'll file a new ticket for 3.16.

[ryandesign (Ryan Carsten Schmidt)]

Thanks; sorry, I missed this ticket. I had previously worked on this update but I stopped working on it when I encountered a build failure:

/usr/bin/clang -arch x86_64 -o gpsdecode gpsdecode.o -L. -lgpsd -lgps_static -lm -ldbus-1
ld: library not found for -ldbus-1

I'm guessing your version attached here fixes that problem by using dbus_export=no.

---

You've changed the qt variant to use qt5, while still including the qt4 portgroup (see the top of the portfile). #47739 suggested that qt4 and qt5 variants be offered.

[fhgwright (Fred Wright)]

Replying to ryandesign@…:

Thanks; sorry, I missed this ticket. I had previously worked on this update but I stopped working on it when I encountered a build failure:

/usr/bin/clang -arch x86_64 -o gpsdecode gpsdecode.o -L. -lgpsd -lgps_static -lm -ldbus-1
ld: library not found for -ldbus-1

I'm guessing your version attached here fixes that problem by using dbus_export=no.

Although I did fix the dbus variant logic, I also got the dbus linking to work by removing a bunch of "LIBPATH='.'" arguments in SConstruct. I'm not sure why those were put in at all (as part of the SConstruct rework from 3.14 to 3.15), since the default LIBPATH is correctly set up to include both ${prefix}/lib and the current directory, so the only effect of those arguments is to remove ${prefix}/lib from the library search path. Of course on most platforms, libraries installed from packages go into the "system" library area, which is searched by default, so the offending arguments are probably a don't care on most platforms.

---

You've changed the qt variant to use qt5, while still including the qt4 portgroup (see the top of the portfile). #47739 suggested that qt4 and qt5 variants be offered.

I saw that in working on the 3.16 port, where I'm currently wrestling with Qt issues right now. :-) My initial thought was to move to qt5 in general, for consistency with the direction things are moving, and the fact that the qt5 and qt4-mac ports claim not to conflict with each other makes a qt4 variant less important. But qt5 refuses to build for PPC (and gives blacklisted compiler warnings on anything older than 10.8), so moving to qt5 completely doesn't work. Meanwhile, the gpsd SConstruct script is using pkg-config to check for the QtNetwork library, but the qt4-mac port seems not to set up the pkgconfig data. I need to answer the "how did this ever work" question before figuring out what to do about it.

As noted above, I've abandoned the 3.15 port and am currently working on 3.16, so this ticket can be closed.

[fhgwright (Fred Wright)]

P.S.: Further discussion should probably move to #50288.

[fhgwright (Fred Wright)]

Could someone with suitable access rights please close this ticket (which is superseded by #50288)? The only "Action" option I have here is "leave as new".

There doesn't seem to be an "obsolete" option for "Resolution, so I guess "wontfix" is the closest.

[g5pw (Aljaž Srebrnič)]

closed as per request.