- Update
- Clean up the master portfile
- Add support for
- python3
- webkit2-gtk3
- Move libproxy-gnome to libproxy-gnome2
- I would like some input on how to do the PYTHON part properly, as it looks rather hacky at the moment.
- libproxy-mozjs is broken on 10.4
Details
Details
- Reviewers
kwm - Group Reviewers
gnome portmgr - Commits
- rP473062: Update net/libproxy to 0.4.15
- exp-run
Diff Detail
Diff Detail
- Repository
- rP FreeBSD ports repository
- Lint
No Lint Coverage - Unit
No Test Coverage - Build Status
Buildable 16999 Build 16867: arc lint + arc unit
Event Timeline
| net/libproxy-python/Makefile | ||
|---|---|---|
| 12 | Python? | |
| net/libproxy/Makefile | ||
| 28 | This is a bit wrong, you define options, and then you say all options are excluded. The current implementation handles this by enforcing the _SLAVE before enforcing the _EXCLUDE but this could change. This should probably not exclude the current set option. | |
Comment Actions
Side note, I don't really like abusing OPTIONS with a vengeance like this, I think it was better before with the LIBPROXY_SLAVE variable.
Comment Actions
No, OPTIONS_EXCLUDE/SLAVE is for when you have "some condition" when a port only supports a subset the the master port options.
For example, most of OPTIONS_EXCLUDE_<arch>=DTRACE, or the remaining ports with a -nox11 slave port where the master port has options.
Here, in the end, none of the ports have options. Thus what I said about really abusing the options framework to gain a few helpers that are not really needed.
| net/libproxy/Makefile | ||
|---|---|---|
| 42–43 | Both these should be WITH_${OPTIONS_SLAVE}. | |
| net/libproxy/Makefile | ||
|---|---|---|
| 42–43 | Mmmm, no, only the first, but the way this is done is so convoluted that it should probably not be done this way. | |
| net/libproxy/Makefile | ||
|---|---|---|
| 42–43 | I see, your not a fan of my cleverness ... I'll revert it to the old way :) | |
Comment Actions
Light visual inspection only, seems mat@ already pointed out all the framework issues.
Also feel free to axe the "$MCom:" lines, those where from the old GNOME svn development repo.
| net/libproxy-kde/Makefile | ||
|---|---|---|
| 5 | Don't forget to reset this one. | |
Comment Actions
Fix CMAKE_ARGS for libproxy-webkit3.
- Now they all should build fine on 10.4@32 and 12.0@64.
Comment Actions
In all the slave ports, the LIBPROXY_SLAVE definition is in a bad place, it should go along the MASTERDIR definition.
Also, I think all the ${.CURDIR:H:H}/net/ might be replaced by ${.CURDIR}/../ as they are all in the net category, I find it more readable.
| net/libproxy-python/Makefile | ||
|---|---|---|
| 15–17 | This should happen before the target. | |
Comment Actions
There are still two .CURDIR:H:H occurrences in the master port. :-)
| net/libproxy/Makefile | ||
|---|---|---|
| 42–43 | Oh, no, I love clever people, but you have to remember that you are not the only one who will work on any given port. If it takes more a minute or two to understand what is happening, it could probably be done in an easier way. In this case, now that I have thought more about these 2 lines (so more like 30 minutes of thinking) looking at it, I think using options, only the _BOOL was needed, because when the options was excluded, it would be "off" and the WITH_FOO=OFF would be added. (I may be wrong, though.) So being clever is nice, but the more clever you think you are, the more you can create bugs that will take longer to figure out, and the more you, yourself will wonder wtf you wanted to do when you did this change 18 months ago. | |
Comment Actions
Mmmm, I had not seen that in the master port, the CURDIR:H were before any include, and thus PORTSDIR did not exist, sorry, you should change back the CURDIR/.. with CURDIR:H :(
Other than that, looks good.
Comment Actions
I was looking at the LIBPROXY_SLAVES variable, and there is two places where you use it directly in its uppercased form, and three dozen where you use the lowercased version, it could probably be simplified by making its content lowercase and use :tu in the two places where an uppercase version is actually needed.
Comment Actions
my guess is your nexst suggestion is to get rid of LIBPROXY_SLAVE and see if PKGNAMESUFFIX is set? :D
Comment Actions
Not sure, I would probably go the other way and set PKGNAMESUFFIX in the master Makefile.