Created attachment 161941 [details] typescript of portmaster run Running: FreeBSD g1-252.catwhisker.org 10.2-STABLE FreeBSD 10.2-STABLE #171 r289170M/289180:1002501: Mon Oct 12 04:05:14 PDT 2015 root@g1-252.catwhisker.org:/common/S1/obj/usr/src/sys/CANARY amd64 with GraphicsMagick-1.3.21,1 currently installed; after updating ports working copy to r399117, "portmaster -ad" found one port to update (graphics/GraphicsMagick), but that failed, claiming 14 test suite failures. I've attached copies of the typescript from running portmaster, as well as the resulting test-suite.log; the port options in effect are: g1-252(10.2-S)[3] make -C /usr/ports/graphics/GraphicsMagick showconfig ===> The following configuration options are available for GraphicsMagick-1.3.22,1: DOCS=on: Build and/or install documentation DPS=on: Display Ghostscript support (must also select X11) FPX=off: FlashPix image format support OPENMP=on: Parallel processing support via OpenMP Q8BIT=off: Use 8-bit pixels (speed) instead of 16 (quality) SSE=off: Use SSE optimized routines TESTS=on: Run bundled self-tests after build WEBP=on: WebP image format support X11=on: X11 (graphics) support ===> Use 'make config' to modify these settings g1-252(10.2-S)[4]
Created attachment 161942 [details] test-suite.log Seems that I wasn't able to add >1 attachment when creating the report.
you know portmaster is unmaintained right? It's pretty much an "use at your own risk" software right now. If it's not updated soon to handle the ports tree it's going to be removed from ports. Given that, I don't expect this PR to get any attention, especially since GraphicsMagick is also unmaintained.
For my machines that install from pre-built packages, I just completed a poudriere run (to build those packages); the build of graphics/GraphicsMagick using poudriere was apparently successful. Thus, perhaps the "bug" actually is in the "unmaintained" (but still documented as "the recommended tool for upgrading installed ports without depending upon other ports" in doc/en_US.ISO8859-1/books/handbook/ports/chapter.xml as of r47483, which is the latest revision of that chapter in the doc repo as of r47601; Last Changed Date: 2015-10-17 03:07:37 -0700 (Sat, 17 Oct 2015)). Oddly, though, an attempt to merely "make -C /usr/ports/graphics/GraphicsMagick fails the same way the portmaster attempt did.
In a post to ports@ (<http://docs.FreeBSD.org/cgi/mid.cgi?20151017203605.31d90511>), the author cited a condition that appears to match the catalyst for this PR, and then mentioned "The fails are gone if the previous version is deleted first." So I tried it on my laptop --- today, running: FreeBSD g1-252.catwhisker.org 10.2-STABLE FreeBSD 10.2-STABLE #176 r289436M/289451:1002501: Sat Oct 17 04:19:41 PDT 2015 root@g1-252.catwhisker.org:/common/S1/obj/usr/src/sys/CANARY amd64 with a ports working copy at r399529, and verified that the following sequence build and installs GraphicsMagick-1.3.22,1 successfully: pkg delete graphics/GraphicsMagick portmaster graphics/GraphicsMagick So I suspect that the tests in question aren't actually testing the just-built version if an old version is currently installed.
Okay. I'll close the PR because your particular issue has been resolved. Obviously there's a bug there but this is well known (I think) so we don't need to carry another PR to track it.