mozilla :: #webdev

16 Mar 2017
17:58ErikRose_rhelmer: Working on :member: option:
17:59rhelmerman it's going to be so much work to beat Firefox code into shape :/ I wonder if post-57 we should go do a big push to remove code that isn't useful, and document the code that we're keeping...
17:59ErikRose_At least you'll only have to do it once, instead of continually copying and pasting onto MDN.
18:01rhelmerErikRose_: "neglecting to copy/paste changes onto MDN" is more like it :P
18:02rhelmeralso tbqh a lot of the code snippets on MDN could stand some review, not just on the technical writing but on correctness/best practice/etc
18:03rhelmerfor internal docs I think optimizing for correctness over better prose is not ideal but OK
18:04rhelmerweb standards should be held to a higher standard
18:11ErikRose_As long as we have an easy way to submit patches :-)
18:27OsmoseIt'd be interesting to see how you could generate a compatibility table from Sphinx markup
18:28ErikRoseProbably with a custom Sphinx directive
18:28OsmoseThere's already some deprecation directives
18:28ErikRoseAssuming all the various routines and selectors and such have "introduced in 57" in them someplace.
18:28rhelmerErikRose_: yeah. mdn is a much lower bar for contribution, having to clone a repo is painful.
18:29ErikRoseIf only we had it on GH.
18:29rhelmerheh, well there are mirrors like gecko-dev
18:29ErikRoseAny writeable?
18:30rhelmerI'm not sure how the syncing happens - I suspect it's one-way from hg->git
18:30ErikRoseI fear so as well.
18:30ErikRosegps would know.
18:30rhelmerbut, two-way syncing is being worked on... rust crates, servo, etc. live on github and need it.
18:30ErikRoseAnyway, a problem for another day
18:31rhelmeryeah. two-way sync to/from gh is something gps is working on for servo/quantum I believe
18:32rhelmerif we can pull it off, I think that'll be a pretty good way to go. then smaller projects can live on github and work as normal, but we'll have a single repo that Firefox and all dependencies ships from
18:39ErikRoseWe'll have to come up with something stronger than a SHA1 hash to tie them together.
18:42rhelmerfair enough. speaking of gps he wrote this after the recent sha1 collision,-sha-1,-and-trusting-version-control/
18:45rhelmerErikRose: I'd like to get at least the AddonManager APIs done as a proof-of-concept before trying to get more people on board for cleaning up our jsdoc and using sphinx-js ... I think it's pretty close. We need to figure out what our private/public concept is going to be (I think @private annotations are probably good enough... we could probably have a jsdoc
18:45rhelmereslint rule that errors if methods marked private are used outside of their module. or something.
18:46rhelmerthere are some pretty cool mozilla-specific eslint rules we should try to push upstream, like balanced-listeners is nice.
18:47ErikRoserhelmer: I'm implementing private-member-skipping as part of :members:.
18:47ErikRoseAnd I'm having a :private-members: flag for including them if you wish.
18:48rhelmerErikRose: good idea
18:48ErikRoseAnd I figured out how to tell getters and setters and normal methods apart.
18:48rhelmeroh nice
18:48ErikRoseAnd I'm implementing :exclude-members: because it's easy.
18:49rhelmersome of these modules are huge, I think automodule is going to be necessary.. it'd be nice if it just had sensible defaults like private members skipped by default etc
18:49ErikRose:members: on autoclass will.
18:50ErikRoseSo yeah, why not.
18:50rhelmeryeah autoclass is probably more the right level than the whole module, given how some of these are implemented with public classes that call methods on private classes
18:50rhelmerI like the idea of having good code samples and prose in the rst, much better than raw jsdoc output
18:50ErikRoseThis closure trick is making a mess of my code.
18:51rhelmerclosure trick for hiding private methods?
18:51ErikRoseNo, in Python, for passing the Sphinx app singleton to the directives
18:51rhelmeroh right
18:52ErikRoseI'm going to do something horrible so I don't have to do something even more horrible.
19:26jgauntI'm trying to use eventListeners to catch ppl doing an experiment breaking the rules to disqualify them - however, it seems like the function I associate with them is being executed when I declare them
19:26jgaunti.e. it *seems like* a keyDown event sends a console.log command before any keys are pressed
19:27jgauntis this expected behavior, or am I confused?
19:35ErikRosejgaunt: Weird! Is there a rogue () sitting around?
19:37jgauntErikRose: I don't *think* so, but I'm mixing my own in the context of a library that uses a lot of listeners itself
19:37ErikRoseBoy, best of luck.
19:37jgauntmore going on in the code than what I've written and understand
19:37jgauntheh, ty :)
19:39mythmonjgaunt: can you narrow down the console.log that is firing?
19:39mythmoncan i see that eventListener?
19:40ErikRoseYou could maybe replace the console object with your own thing and set a breakpoint in it. ;-)
19:41mythmonit should be possible to just look at the message logged, and grep the code for that
19:48jgauntErikRose, mythmon - I just vidyo'd with kamyar
19:48jgauntI was calling a parameterized function from my listeners
19:48jgauntrather than anon functions
19:49jgaunttrying the fix, idk what to expect
19:52jgauntnow that my function calls are wrapped in anon functions it works as I expected
19:52jgauntI was foggy on his explanation but something about the listener passing an object to the function implicitly
19:53jgauntwhen the event occurs
19:53jgauntdoing a function call inhibits that somehow
19:53jgauntI am probably doing the description injustice
19:54mythmonjgaunt: can i see the code before and after?
19:55jgaunt^mythmon /me shrugs
19:56mythmonyeah, i'm not sure what Kamyar was trying to explain, but this is exactly what ErikRose thought it was
19:56jgauntthe "rogue ()" ?
19:56mythmonon a line like this, you call disqualify immediately, and then bind its return value as the listener. window.addEventListener('blur', disqualify("Loss of window focus"), false);
19:57ErikRoseYep, there ya go. You passed it the result of a function, not a function itself.
19:57mythmonwhereas with the wrapped function() { ... }, you create a function that is *not* evaluated immediately
19:57jgauntnow there's a succinct explanation, ErikRose
19:57jgauntokay, ty for explaining from another perspective
19:58ErikRosehaha :-)
21:44ErikRoserhelmer: Tada: :members: option.
21:44ErikRoseIt doesn't yet respect privacy or take lists of members or take exclusions, but it works and makes getters and setters and instance vars look nice as well.
17 Mar 2017
No messages
Last message: 8 days and 18 hours ago