Created attachment 155261 [details] deskutils_ganttproject.diff
Please do not abuse PORTDATA. It it not the same as PORTDOCS or PORTEXAMPLES.
As far as I can see, there is no contradiction between the differnce of PORTDATA from others and how it's used in this PR. I have no idea what the purpose of PORTDATA is if it's incorrect.
Created attachment 156706 [details] deskutils_ganttproject.diff Chase r386097
I agree this is an abuse of PORTDATA. Actually, never use PORTDATA. The entire mechanism needs to be removed. Can you fix and resubmit?
It is very difficult for me to guess why the existence of PORTDATA has to be a target for criticism. Except for the ability to prevent installation, there does not seem any difference between DATADIR and DOCSDIR/EXAMPLESDIR about dynamic creation of contents list.
"Why" is not important. What is important is that there is extreme prejudice against PORTDATA, to the point that's its use is considered a bug. Yes, the feature should be removed. I think it probably will be at some point. Thus, don't increase it's used because that means somebody has to revert it later.
(In reply to tkato432 from comment #5) > It is very difficult for me to guess why the existence of PORTDATA has to be a target for criticism. Understanding it probably requires learning some of the history behind these knobs. Originally we had none of them, and provided that docs and examples are often optional, it typically had required setting PLIST_SUB in the Makefile and clumsy lines in pkg-plist: > %%PORTDOCS%%%%DOCSDIR%%some_docfile.txt > ... Docs are usually not just optional and thus require special support in Makefile/pkg-plist, they often are of little interest to grep pkg-plist for, so eventually someone proposed the idea of PORTDOCS knob which allows one to list all the docfiles in a variable and thus remove explicit support bits thereof from the Makefile and pkg-plist. (Rationale for PORTEXAMPLES is pretty much the same.) I'm not sure why PORTDATA was ever created in the same fashion as it is not the same at all. PORTDATA is essential to port's operation, so it should never be installed conditionally (there is one special case when port data assets are huge which is common for e.g. games, but in this case it's usually a better idea to split gamefoo and gamefoo-data ports). As it does not have to be installed conditionally, there's no PLIST_SUBS dances and %%-prefixed lines in pkg-plist to worry about, and last but not least, port data are a lot more interesting to grep for. That said, PORTDATA knob (particularly when used with globbing) should just die, and I prefer sooner than later.
Created attachment 162079 [details] deskutils_ganttproject.diff Version 2.7.1 has been released.
The argument would be persuasive if making use of something being controversial should be avoided for the time being. Anyway, this place does not seem appropriate to discuss the validity of PORTDATA itself. I would like to continue using static plist for this port at the moment.
I'll try it out.
A commit references this bug: Author: marino Date: Fri Oct 16 13:53:51 UTC 2015 New revision: 399475 URL: https://svnweb.freebsd.org/changeset/ports/399475 Log: deskutils/ganttproject: Upgrade version 2.6.2 => 2.7.1 PR: 199223 Submitted by: ports fury Changes: head/deskutils/ganttproject/Makefile head/deskutils/ganttproject/distinfo head/deskutils/ganttproject/pkg-descr head/deskutils/ganttproject/pkg-plist
A commit references this bug: Author: marino Date: Fri Oct 16 14:00:32 UTC 2015 New revision: 399476 URL: https://svnweb.freebsd.org/changeset/ports/399476 Log: deskutils/ganttproject: Unmask installation commands I missed that install commands got masked; unmask them again. PR: 199223 Changes: head/deskutils/ganttproject/Makefile