04:50
<MikeSmith>
heh https://groups.google.com/a/chromium.org/d/msg/chromium-dev/0wWoRRhTA_E/-NxrcXoqBwAJ
04:54
<MikeSmith>
some should reply with a message asking him to estimate what the costs are of making a business decision to tie your entire business model to a single-vendor proprietary runtime instead of betting on the Web instead
05:16
<Uu>
Hi when will the next htm5 specifications public?
05:17
<tantek>
Uu, today: http://html.spec.whatwg.org/
11:03
<FND>
hi - using `fetch`, if a CORS request fails with `fetch`, that failure has to be handled via `catch`
11:03
<FND>
however, in there you don't have access to the server response - which means that you can't be certain what went wrong, exactly
11:04
<annevk>
FND: that is somewhat intentional, the developer console should give you more information though
11:04
<FND>
yeah, I know, but:
11:04
<FND>
my current scenario is accessing SharePoint with an expired access token - SharePoint does provide a `x-ms-diagnostics` header along with a 401 status code
11:05
<FND>
but I have no way to determine that programmatically, so I can't just say "oh, right, I'll just refresh my token before retrying"
11:06
<annevk>
Ideally SharePoint would have CORS on the 401
11:06
<annevk>
Then it would be exposed to you
11:06
<annevk>
So I think if SharePoint provides a CORS access point, this is a bug in their system
11:06
<FND>
good point, actually
11:06
<FND>
(I'm not surprised, their REST API is pretty terrible)
18:32
<annevk>
https://twitter.com/simevidas/status/759080906455969792 is pretty cool
18:33
<annevk>
The word diff tweet this is response to, that is
19:00
<TabAtkins>
annevk: The term "an unclosed node of a node" is extremely confusing to work with. :/
19:02
<TabAtkins>
I think it's because in the construction "ADJ node of a node", we expect ADJ to be a relationship between the two nodes (like "ancestor"), but "unclosed" reads as a personal property of the first node, not a relationship.
19:09
<TabAtkins>
annevk: Maybe "unclosed relative of a node"? It took me some thinking to decipher that it was capturing the notion "not trapped in a closed shadow root in a way that makes it invisible to B".
19:31
<TabAtkins>
annevk: Or a longer phrase "A is potentially reachable from B"
20:05
<jsbell>
smaug_____: It looks like your directory upload FileEntry/DirectoryEntry impl is including some of the useless bits: Entry.filesystem, DirectoryReader.removeRecursively(), FileEntry.createWriter(), FileSystemFlags dict and the DOMFileSystem interface. Have you determined that these are necessary for compat? Edge wasn't planning to implement those. I was hoping we could remove them.
20:22
<jsbell>
smaug____: https://github.com/WICG/entries-api/issues/4 for discussion. Looks like Edge will have Entry.filesystem
20:57
<annevk>
TabAtkins: file an issue? Seems fixable
20:57
<TabAtkins>
kk, will do
20:57
<annevk>
Ta
21:02
<smaug____>
jsbell: also Gecko
21:04
<smaug____>
jsbell: FWIW, we got a list from MS what they were going to implement. (but they then implemented some different set of properties and methods after all)
21:04
<smaug____>
jsbell: also, I'm planning to enable the API in Nightly today or tomorrow
21:19
<jsbell>
smaug____: thanks - as you saw in the bug, I've added the common subset between Blink/Edge/Gecko. Do let me know if anything else seems necessary and I'll happily add/fix.
21:32
<smaug____>
jsbell: FWIW, you can use my real name, Olli Pettay
21:33
<jsbell>
smaug____: Will do.