Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 02:03
    kitsonk commented #3727
  • 02:01
    kitsonk commented #3696
  • Jan 22 23:54
    ry closed #3744
  • Jan 22 23:35
    Qws commented #3754
  • Jan 22 23:31
    nayeemrmn commented #3754
  • Jan 22 23:27
    Qws opened #3754
  • Jan 22 23:17

    denobot on gh-pages

    Update benchmarks (compare)

  • Jan 22 22:58

    bartlomieju on master

    refactor: snapshotting (#3753) (compare)

  • Jan 22 22:58
    bartlomieju closed #3753
  • Jan 22 22:45
    kitsonk commented #986
  • Jan 22 22:40
    kitsonk commented #1158
  • Jan 22 22:28
    kitsonk commented #1421
  • Jan 22 22:22
    kitsonk commented #3751
  • Jan 22 22:08
    bartlomieju synchronize #3753
  • Jan 22 21:46
    bartlomieju synchronize #3753
  • Jan 22 21:18
    bartlomieju review_requested #3753
  • Jan 22 20:58
    bartlomieju synchronize #3753
  • Jan 22 20:57
    ry closed #3123
  • Jan 22 20:57
    ry commented #3123
  • Jan 22 20:45
    bartlomieju synchronize #3753
Kurt Mackey
@mrkurt
neat! I'm only vaguely familiar with aries agents but that sounds pretty intense
Ajay Jadhav
@jadhavajay
Yup, it is..
Its part of the emerging SSI (Self-Sovereign Identity) ecosystem, where agents are useful to manage the Decentralized Identities and their interactions..
Nick Stott
@nstott

I had a gross problem with the tests hanging on that last pr,
the std tests would run, and then the deno process would just hang for ages after the tests had finished, and wait until a socket timed out, at which point the deno process would exit

adding an explicit close(rid) after I get the EOF seemed to fix that, and also seemed to fix denoland/deno#2923

Bartek Iwańczuk
@bartlomieju
@nstott I think that's expected. We can't close resources automatically
Nick Stott
@nstott
it makes sense, it was just a little surprising to me.
Andy Hayden
@hayd
@ry is CPATH the wrong thing here: https://gist.github.com/hayd/c7b2531cf68b22d431490478b221e759
(clang and libxx are both version 9... :/ )
Ryan Dahl
@ry
@hayd No it's not a CPATH issue. It looks like you're using the wrong version of c++ standard (like what flag is being used --std=c++17) ... or maybe it's the compiler you're using. v8 needs a very specific version of clang++
i see your gn args at the top now...
@hayd The version we use is 10
> third_party/v8/third_party/llvm-build/Release+Asserts/bin/clang --version
clang version 10.0.0 (https://github.com/llvm/llvm-project/ 64a362e7216a43e3ad44e50a89265e72aeb14294)
Target: x86_64-apple-darwin18.7.0
Thread model: posix
InstalledDir: /Users/rld/src/rusty_v8/third_party/v8/third_party/llvm-build/Release+Asserts/bin
Andy Hayden
@hayd
hmmm 10 isn't available from downloads yet http://releases.llvm.org/download.html#9.0.0
Ryan Dahl
@ry
it's quite annoying that they rely on the bleeding edge clang
then again - we're running bleeding edge v8
Christian Moritz
@chrmoritz
However llvm/clang 9.0.0 should be enough to build v8 if you set treat_warnings_as_errors=false(as you do) because of some renamed warnings. At least it works for us at homebrew (the system clang on Catalina works too).
Might be worth to try building deno 0.17.0 as a first step, because it's v8 has much lower requirements for the compiler.
Christian Moritz
@chrmoritz
@hayd are you using the libc++ from llvm 9.0.0 too? I might be required to put it in the same prefix as clang (the one you pass to clang_base_path=).
Andy Hayden
@hayd
@chrmoritz It's this Dockerfile https://gist.github.com/hayd/e6d3efbf1bc2fcecbd0ef50e3245e928#file-dockerfile-L39-L46
(trying to build for Amazon Linux 1)
I am thinking I am pointing to the wrong libc++, but not sure what it should be. It isn't in the same prefix!
Andy Hayden
@hayd
On Alpine Linux I was able to get much further, and that was with clang 9.0.0.
Christian Moritz
@chrmoritz
@hayd You need to set your clang_base_path="/tmp/clang+llvm-9.0.0-x86_64-linux-sles11.3"
Christian Moritz
@chrmoritz
Also the std::is_trivially_copyable is the same as in nodejs/node#30020, fixed by https://chromium-review.googlesource.com/c/v8/v8/+/1868622 ?
Andy Hayden
@hayd

@chrmoritz I think I'd tried that permutation, does seem to hit that: https://gist.github.com/hayd/bf1abf2553dc328049313ec22cf91cf4
I guess that fix isn't in deno's v8 yet: https://github.com/denoland/deno_third_party/blob/master/v8/src/base/macros.h

So, the problem is deno isn't bleeding edge enough :laughing:

Andy Hayden
@hayd
Should it be using libc++ from third_party? Maybe that was the issue?
https://gist.github.com/hayd/aeaef216a264644f4a88a78650589539
(I've no idea whether this is further progress or backwards. )
Vincent LE GOFF
@zekth
@ry you want me or kevin to archive denolib/node and merge it in deno_std under 'node' folder? We also need to update the documentation of std and maybe write the list of remaining modules to port
Ryan Dahl
@ry
@zekth Regarding std/node - yes - please
@zekth Regarding documentation - have you seen that we have a jsdoc viewer now? https://deno.land/std/http/server.ts?doc
@zekth That makes it a bit more fun to do documentation PRs
Christian Moritz
@chrmoritz
@hayd https://gist.github.com/chrmoritz/b9a1a73722c3591e244e6d9aa6ca84bc workd for me without errors using Googles custom libcxx
Stanislau
@venikman
Hi all! I am trying to install deno on Pi Zero, look like it doesn't work. I installed raspbian-buster on it and I am getting Unsupported architecture armv6l. Only x64 binaries are available. . Is there is any workarounds to install deno in my small pi ?
I found issue (denoland/deno#2295) about it but may be found hacks to make it work?
Christian Moritz
@chrmoritz
@venikman Unfortunately there are no official arm binaries available yet. So you would need to compile it from source by yourself.
Andy Hayden
@hayd
@venikman You can also run it using docker (and can run docker from raspbian-buster).
Vincent LE GOFF
@zekth
@ry nice! I'll check it when i get back from asia in 1 week
Andy Hayden
@hayd

To announce to the channel, in case people aren't following #1456, I was able to get deno 0.22.0 working in AWS Lambda. An interesting part is you can pass the DENO_DIR to avoid any runtime compilation. https://github.com/hayd/deno_docker/tree/master/aws-lambda#prepare-zip-file-to-upload-to-aws-lambda

Still a bit to flesh out... also not sure on best practices (e.g. layers / serverless).

Daniel Buckmaster
@crabmusket
DENO_DIR manipulation is super helpful. I'm using the same trick to ship a Docker container containing the entire application source
between that, import maps, and the new lock file, Deno is shaping up to be a great application platform as well as a great scripting/learning platform! I think "import from URL" would be an awesome resource for anyone teaching programming; not having to explain package managers, etc etc
Daniel Buckmaster
@crabmusket
has there been any discussion of how to access private code in Deno? e.g. I have a private repo at gitlab.com/company/app, and another private repo at gitlab.com/company/library. how can app import from library by URL? is that what HTTP proxies are for? is there a recommended solution aside from a monorepo? just using import maps to a vendored copy?
Bartek Iwańczuk
@bartlomieju
I believe there was an issue asking for support for authorization headers
Daniel Buckmaster
@crabmusket
using git submodules along with import maps could also be a potential solution, if users are expected to have clone access to the repo. if they're not, then the code is probably too well-protected :p
it seems like there might be a lot of potential solutions to the problem of "how to work with private code" - does this warrant an FAQ in the docs?
Andy Hayden
@hayd
Andy Hayden
@hayd
@crabmusket submodules are a bit annoying, specifically it means there's multiple places to update (not just deps.ts), but on the other hand it means all the code is there...¯_(ツ)_/¯
Daniel Buckmaster
@crabmusket
I agree that submodules are annoying!
Kevin (Kun) "Kassimo" Qian
@kevinkassimo
Is there a way to prevent TS compiler from removing unused imports in compiled code?
This would render lock useless until I actually use stuff imported from remote module
Ghost
@ghost~548765fcdb8155e6700dd7bf
Anyone here on Telegram? If so, please join GNU Propoganda if you love free software.
@GNUPropaganda
And please upgrade to Spectrum chat or IRC. I had to agree to a ToS just to speak freely work you guys (and Glas, hopefully).
gals*
Kitson Kelly
@kitsonk

Is there a way to prevent TS compiler from removing unused imports in compiled code?
This would render lock useless until I actually use stuff imported from remote module

There is a long standing issue of this, if you don't want it to elide, just use import "foo.ts"... If you are using a type only imports, you need to do it twice then. Lots of people complain about it, but I side with TS team, if you only need types from the file, the import should be erased, just like types are erased. @kevinkassimo