Maxim Cournoyer writes: >> I did not found a way to report a bug. Nothing is listed in the manual, >> and their apcupsd-users mailing list requires JavaScript to sign up (and >> does not tell me the email address otherwise). >> >> I did try to blindly send an email, will see if it will be accepted >> without being subscribed. If that happens, I will add the link here. > > OK, thank you. I have now added a link to my message to the mailing list. >>> >>>> + ;; Enable additional drivers. >>>> + "--enable-test" >>> >>> Is '--enable-test' useful? What does it do? >> >> According to the manual >> >> This turns on a test driver that is used only for debugging. >> >> >> Since there does not seem to be any harm in having it on, I enabled it, >> since Guix seems to aim for feature complete packages (I assume that is >> the reason for everything being so large). But I have no use for it, so >> I can turn it off if you would prefer. > > For this kind of very edge case that is not enabled by default, I'd > leave it off. Our maximalist take on things doesn't including switching > on every non-default option, but if extra dependencies are automatically > picked as part of the configure script, then we would typically add > them, unless they grow the 'guix size' of the package unreasonably (like > adding libreoffice to emacs-org-mode because it has support for it :-)). > > It's one of the reason. Other reasons are references kept erroneously, > static libraries, large documentation, wrappers keeping references to > native inputs that shouldn't be captured, etc. There's a lot of work to > do to reduce the size of the distribution :-). > > [...] Makes sense, I took it out. >>>> + (add-before 'configure 'touch-txt-docs >>>> + (lambda _ >>>> + (for-each (lambda (f) >>>> + (call-with-output-file f close-port)) >>>> + '("doc/apcupsd.man.txt" >>>> + "doc/apcaccess.man.txt" >>>> + "doc/apctest.man.txt" >>>> + "doc/apccontrol.man.txt" >>>> + "doc/apcupsd.conf.man.txt")))) >>> >>> I think I'd rather depend on these than introduce hacks like below. >> >> The "hacks" *below* would still be required. The HTML manual is not >> built nor installed by default, and there is no target to invoke to >> install them. So both 'build-manual and 'move-doc phases would still be >> required. >> >> We could get rid of the "hack" *above* ('touch-txt-docs), true. It is >> worth those two additional inputs? > > Yes, I meant above, sorry. I don't see mandoc and util-linux as large > dependencies enough to justify working around these, especially if they > are only used at build time. And, the would actually be useful when > working on the source of acupsd in a 'guix shell -D acupsd' environment, > so I'd just add them. > > [...] > Makes sense, I will remove the phase and add the dependencies into native-inputs. >>> Could you please send a v2? >> >> Definitely, I will go through your review for the service file as well, >> and once I have all the answers (both here and there), will send v2. > > I haven't seen v2 yet, I guess you are still working out the service > part? I did not get to the service part yet (life has sadly been somewhat busy), will get to it during this weekend. But I was also waiting for your response here, there was no reason to send v2 since I was pretty sure I will need to do additional changes, so v3 would required anyway. Tomas -- There are only two hard things in Computer Science: cache invalidation, naming things and off-by-one errors.