These are chat archives for assetgraph/assetgraph

26th
Sep 2016
Peter Müller
@Munter
Sep 26 2016 05:04
Nice!
I'm just wondering why it never worked on my machine
Andreas Lind
@papandreou
Sep 26 2016 15:10
@Munter I know this one is ridiculous to review: assetgraph/assetgraph-builder#454 ... if you agree with the overall direction of it, we can just merge it and fix any breakage later.
Peter Müller
@Munter
Sep 26 2016 15:12
This is certainly true while in on a phone :D
Andreas Lind
@papandreou
Sep 26 2016 15:12
;)
Peter Müller
@Munter
Sep 26 2016 15:14
By the way the canonical branch now passes all tests. We just need to find a way to not use hreftype of you still fell strongly about that
Andreas Lind
@papandreou
Sep 26 2016 15:16
@Munter I do (I hope I've managed to explain why as well :)
Andreas Lind
@papandreou
Sep 26 2016 15:22
I guess my main point is that it doesn't capture protocol-relative urls pointing at the canonical root, so we'd need both a canonicalAbsolute and canonicalProtocolRelative href type
... which suggests that it's an orthogonal thing
Peter Müller
@Munter
Sep 26 2016 15:28
I was working under the assumption of only exact matches
Andreas Lind
@papandreou
Sep 26 2016 15:28
@Munter It's certainly better than what we have now.
But it would be really nice to solve it in the general case.
Peter Müller
@Munter
Sep 26 2016 15:52
Do we need an isCanonical getter to describe this instead? Then the hreftype still makes sense without the extra overhead
Then we can choose to support greater breadth of matching to the canonical root later
Andreas Lind
@papandreou
Sep 26 2016 15:54
@Munter That's one possibility. Or if it turns out that we never really need isCanonical in practice, we can also just build it into the crossorigin getter so it returns false for absolute and protocol-relative links that point into the canonical root?
Peter Müller
@Munter
Sep 26 2016 16:05
I think it's better separate. To keep the intention clear. Especially when it grows from absolute match to source expression match implementation
Also, ability to query for it
The only thing we lose from the hreftype is the ability to switch it around. That could also be done with a setter I think
Andreas Lind
@papandreou
Sep 26 2016 16:29
You're probably right. Will be interesting to see how it looks in buildProduction.
Munawwar
@Munawwar
Sep 26 2016 17:36
Just tried ag-b on windows's "linux subsystem"..
Munter @Munter crosses fingers
Munawwar
@Munawwar
Sep 26 2016 17:37
fails to run buildProduction on some dependency of postcss
called macaddress. os.networkInterfaces() <- function undefined
But it's awesome that it came that far
Peter Müller
@Munter
Sep 26 2016 17:38
So it installed without an error and then failed somewhere during the run?
Munawwar
@Munawwar
Sep 26 2016 17:39
Kind of. There was some error on one of the optional depenency though.
But that doesn't seem to be related.
Peter Müller
@Munter
Sep 26 2016 17:40
Those are mostly optional for exactly that reason. Hard to install and with no blocking functionality. Not key
It's pretty cool it installs
Christina just bought a windows machine. What do I need to do to make it able to run linux subsystem?
Andreas Lind
@papandreou
Sep 26 2016 17:43
Microsoft/BashOnWindows#468
Munawwar
@Munawwar
Sep 26 2016 17:43
First she needs to install the HUGE windows update
Yup...os.networkInterfaces() seems to throw errors for me.
Munawwar
@Munawwar
Sep 26 2016 17:54
I wonder why macaddress is needed anyway :/
assetgraph/node_modules/cssnano-paras20xx/node_modules/postcss-filter-plugins/node_modules/uniqid/node_modules/macaddress/index.js
Ok, I answered my own question - https://www.npmjs.com/package/uniqid. Still...like why!?
Peter Müller
@Munter
Sep 26 2016 19:11
Looks like it shoul dbe an optional dependency: https://github.com/bayorigin/uniqid/blob/master/index.js
Andreas Lind
@papandreou
Sep 26 2016 21:15
It's probably only treated as optional when running in the browser.