| 00:01 | <SamB> | the one thing I've liked about it in the past is that I can actually get more-or-less a whole stack trace for code compiled with it |
| 00:01 | <SamB> | but maybe I could do that with an up-to-date GDB, too |
| 00:03 | <SamB> | (with GCC on windows, I've had to resort to stuff like using WinDbg alongside GDB and/or binutils) |
| 00:04 | <SamB> | (where WinDbg can unwind to my code -- and very likely also through it -- and GDB or binutils can help me with the addresses/name mappings) |
| 00:07 | <SamB> | (hmm, I guess intelisense is also kind of cool ...) |
| 00:07 | <SamB> | (but not cool enough to really want to deal with cl.exe ...) |
| 00:08 | <SamB> | (or, well, the compiler proper's probably in DLLs now ...) |
| 00:09 | <SamB> | Hixie: isn't the outline for http://www.whatwg.org/specs/web-apps/current-work/multipage/the-end.html kind of nonsensical? |
| 00:11 | <Philip`> | zewt: "#define _CRT_SECURE_NO_WARNINGS" has worked in all the cases I've seen so far |
| 00:12 | <SamB> | do they seriously have a read_s? |
| 00:13 | SamB | reported https://www.w3.org/Bugs/Public/show_bug.cgi?id=26012 |
| 00:15 | gsnedders | finally has a moment in which he thinks having text nodes is a good distinction from strings |
| 01:16 | <Domenic> | Hah, I like how in the WebKit source code there's an actual callTheAdoptionAgency(token); line |
| 01:19 | <SamB> | Domenic: well how would YOU name something that implements "call the adoption agency"? |
| 01:50 | <caitp> | "activateTerribleDesignDecision()" |
| 02:01 | <caitp> | seriously, who ever came up with that reparenting scheme... SMH ._. |
| 02:37 | <gsnedders> | caitp: see the acknowledgements section of the spec; Dave Hyatt |
| 02:37 | <gsnedders> | caitp: and if you start with the requirement that the DOM must be a tree, it's not clear you can do much better :( |
| 02:38 | <caitp> | gsnedders, this discussion has been had with hixie --- the DOM doesn't break when things don't get reparented, you can see this by dynamically inserting arbitrary content at places where it would ordinarily be reparented during parsing |
| 02:39 | <caitp> | it was a mistake to ever decide the DOM needed to reparent things, and it turns out to be very hard to fix that mistake ._. |
| 02:39 | <caitp> | but I don't like to name names when mocking terrible design decisions |
| 02:40 | <caitp> | people did it wrong before, and I guess we're stuck with it forever :( |
| 02:40 | <gsnedders> | But if you don't reparent you don't get a DOM that renders correctly, no? |
| 02:41 | <caitp> | you get a DOM that renders the way you expected it to be rendered when you authored your content wrong |
| 02:42 | <caitp> | the reparenting thing A) complicates things for no good reason, and B) makes things like custom elements slightly harder to implement correctly without the use of hackish nonsense like is=* |
| 02:43 | <caitp> | all to avoid breaking some awful smut website, I mean really |
| 02:43 | <gsnedders> | Plenty would argue "to avoid breaking some website" is a good reason. :) |
| 02:43 | <caitp> | people were not thinking when they decided this was something that needed to happen, but it's too hard to fix it =( |
| 02:46 | <gsnedders> | But, eh, the ship sailed when it came to misnested content tags before I could speak (probably). |
| 02:48 | <caitp> | from my perspective, my generation shouldn't have to pay for mistakes made by the generation that came a decade or two before, we should be fixing their broken crap |
| 02:48 | <caitp> | but herr ian disagrees |
| 02:49 | <gsnedders> | meh, plenty of evidence that breaking websites leads web browsers to lose marketshare and not websites being fixed |
| 02:50 | <caitp> | from my perspective, that's not a real excuse, if it leads to future work having to satisfy earlier, broken work |
| 02:50 | <caitp> | but people have their priorities, I guess |
| 02:50 | <caitp> | I just think they're getting it wrong in this case |
| 02:50 | <gsnedders> | convince the browser vendors to lose their marketshare, I guess |
| 02:50 | <gsnedders> | And there's again plenty of evidence what's in a spec doesn't really effect what browsers do in terms of breaking sites |
| 02:51 | <caitp> | that is certainly true |
| 02:51 | <caitp> | but the fact that the reparenting scheme lives in the spec sort of validates the decision to include broken stuff in the browser |
| 02:52 | <gsnedders> | Only difference the spec dropping it would make is the spec would cease to have implementations |
| 02:53 | <caitp> | if all vendors decided to ignore it |
| 02:53 | <gsnedders> | which they will |
| 02:53 | <caitp> | I don't think you can say "they will" |
| 02:53 | <caitp> | there are a lot of implementations out there |
| 02:54 | <gsnedders> | well, any impl with any real number of users will ignore it |
| 02:55 | <caitp> | I'm not sure about that, I don't think fixing that particular part of brokenness would hurt marketshare that much for a mainstream browser |
| 02:55 | <caitp> | and it would be temporary |
| 02:56 | <caitp> | you know why it's temporary, because people have their browser of choice, it's part of their identity, it becomes who they are! |
| 02:56 | <caitp> | that sounds silly, but it's not that much of an exaggeration, people have preferences for this stuff |
| 02:56 | <caitp> | pretty strong, irrational preferences |
| 04:21 | Hixie | runs into the fact that the stack of open elements is sometimes randomly mutated |
| 04:22 | <Hixie> | bummer |
| 04:22 | <Hixie> | that's gonna be exciting. |
| 05:18 | <Hixie> | hey someone double-check me here. In "push onto the list of active formatting elements", where the spec says "list markers", it should say "scope markers", no? |
| 09:19 | <jgraham> | Hixie: AFAICT there you use "list marker", in the definition above you use "scope marker" and everywhere else you use "marker" |
| 16:08 | <Hixie> | jgraham: k, that's what i concluded too |
| 16:15 | <jgraham> | Hixie: FWIW I think "scope marker" is a bad name since it requires a note that it doesn't correspond to the other notion of "in scope" |
| 16:41 | <Hixie> | jgraham: yeah... |
| 16:46 | <Hixie> | ok, renamed them all to just "marker", that's unambiguous enough |
| 16:46 | <Hixie> | also, made them all cross-refs |
| 16:50 | <jgraham> | Hixie++ |
| 23:50 | <gsnedders> | http://anolis.gsnedders.com/ seems pretty silly |
| 23:50 | <gsnedders> | Why doesn't that just redirect? |
| 23:50 | <gsnedders> | o_O |