Simon Wright <[email protected]> writes: > On 4 Jul 2016, at 14:28, Stephen Leake <[email protected]> wrote: >> >> The current org.emacs.ada-mode head, >> 36709c58de7c7886d6baaa366aeedcfa1417644a, is a release candidate. It >> passes all my tests on Windows. >> >> If no one identifies any problems by next Friday (8 July), I'll release >> it. > > build/Makefile target "install" doesn't handle Darwin properly - could > modify it to understand that Darwin doesn't use .exe, as per attacked > patch, or could go the gprinstall route.
gprinstall makes sense. > I also had Darwin-related problems checking out the elpa stuff (no > reply yet on my application): I don't have the arm-info stuff > installed, but more tricky is that the elpa GNUmakefile uses > linux-only switches to tar (they may well be GNU tar switches, of > course). Ok. I think this means you can't do the official release, but you are working on getting to the point where you could. > What has to change to up the version to 5.1.10? As far as elpa is concerned, just change the Version header in ada-mode.el. However, that version number is duplicated in several places; there is a list in ada-mode/notes.text, at the end of the file under "release process". > With regard to gpr_query, I think there is a problem in that both > ada-gnat-xref.el and gpr-query.el 'provide' ada-xref-tool. > I get the strong impression that a feature should be defined in (at > most) one file? Right; they are alternate implementations, only one should be loaded. Hmm. That rules out multiple active projects, some of which use 'gnat, some use 'gpr_query. Which we want to support. > The result is that (if you're not using xref-tool in a .prj, anyway) > you always end up with 'gnat. Hmm. I don't; I always end up with 'gpr_query. But I think we should get rid of the 'ada-xref-tool feature, and just use the variable, which can then change value at run-time. > See attached patch. your next email looks good, but also set ada-xref-tool (see below). > On this topic, the handling of ada-xref-tool is different in the two files > that set it? > > ada-xref-tool: Actually ada-gnat-xref.el > > (unless (default-value 'ada-xref-tool) > (set-default 'ada-xref-tool 'gnat)) This assumes someone might set ada-xref-tool to be buffer-local, which is not the intended use. I always set variables buffer-local at their declaration if that is the intended use (for some reason, that is not the common Emacs style). I was probably experimenting with making ada-xref-tool buffer local, but then gave up on that. I think we should not set ada-xref-tool here at all; just set it in ada-mode.el if it is not already set. That makes it clearer that ada-gnat-xref.el and gpr-query.el can both be loaded at the same time. > > gpr-query.el: > (unless (and (boundp 'ada-xref-tool) > (default-value 'ada-xref-tool)) > (setq ada-xref-tool 'gpr_query)) This is inconsistent; it checks the default value, but sets the (possibly) buffer local value. Just delete it. Please make the changes discussed here, since you can test the makefile change on Darwin. -- -- Stephe _______________________________________________ Emacs-ada-mode mailing list [email protected] http://host114.hostmonster.com/mailman/listinfo/emacs-ada-mode_stephe-leake.org
