Hello, Try removing the call to dh_auto_build from your override_dh_auto_build. It's doing a bunch of byte compilation that is unnecessary because dh-elpa will do that later on.
On Tue, Jun 28, 2016 at 02:01:41PM +0300, Dmitry Bogatov wrote: > In general, do we have any script, that automates process of writing > watch files? Every time I need to write watch file for github, I > copy-and-paste from comments in /usr/bin/uscan. There is the uscan manpage, rather than looking at the code, which you might consider slightly better. On Tue, Jun 28, 2016 at 04:49:33PM +0300, Dmitry Bogatov wrote: > Do not understand. I alread have tarball content, as I downloaded > it at 'upstream/1.2.12'. What more we need? It just means that your tarball will be different from the one that actually gets uploaded by the DD who sponsors the RFS. Hopefully it is not materially different (just timestamps etc.) but I think that settings in ~/.gbp.conf could mean a materially different tarball gets generated on the DD's machine. It's a nice safety check if we can check out your tarball from the pristine-tar branch. But not a big deal and certainly doesn't block sponsorship. > Here is script, that does same as dh_elpa_test: > > emacs -batch -Q -L . --eval "(require 'evil)" -l package \ > --eval "(add-to-list 'package-directory-list > \"/usr/share/emacs/site-lisp/elpa\")" \ > --eval "(add-to-list 'package-directory-list > \"/usr/share/emacs/site-lisp/elpa-src\")" \ > -f package-initialize \ > -L . \ > -l evil-tests.el \ > -l lib/ert.el \ > -L lib \ > -f evil-tests-initialize > > It reports 22 failures. If I replace -batch with -nw, all tests > passes. So seems tty is really needed, but I do not understand why. This is not what dh_elpa_test is actually running ;) It's calling `ert-batch-run-tests-and-exit' instead of `evil-tests-initialize'. Before we can debug this properly, we need to make dh_elpa_test do the same thing that upstream's makefile does (except the -nw/-batch). Please investigate DH_ELPA_TEST_ERT_RUNNER (hint: add debian/run-tests.el). In the worst case, if we can confirm that some of tests really do require a tty (it would be good to figure out why -- maybe ask upstream?), you can add a patch disabling those tests in particular. > > It seems it wasn't enough. If I move my .emacs.d out of the way and > > then run it, and M-x evil-mode, I get this: > > > > Error in post-command-hook (evil-repeat-post-hook): (void-function > > evil-repeat-post-hook) > > Error in pre-command-hook (evil-repeat-pre-hook): (void-function > > evil-repeat-pre-hook) > > Patched it. Check again. Nice work. Have you forwarded the fix upstream? -- Sean Whitton
