https://pantsbuild.org/ logo
Powered by
#general
  • e

    enough-analyst-54434

    08/27/2015, 2:08 AM
    aha its an optional dep
  • e

    enough-analyst-54434

    08/27/2015, 2:23 AM
    oooook, I'll find another example: ```
  • e

    enough-analyst-54434

    08/27/2015, 2:23 AM
    ```a
  • e

    enough-analyst-54434

    08/27/2015, 2:24 AM
    I do not understand slack code fencing
  • e

    enough-analyst-54434

    08/27/2015, 2:24 AM
    So this: ```$ rm -rf * && npm install webpack fsevents -g --prefix ./webpack-canonical --cache ./npm-cache --verbose
  • e

    enough-analyst-54434

    08/27/2015, 2:24 AM
    @#$!
  • e

    enough-analyst-54434

    08/27/2015, 2:24 AM
    At any rate, fsevents is OSX only:
    Copy code
    npm ERR! Linux 4.1.6-1-ARCH
npm ERR! argv "/home/jsirois/.nvm/versions/node/v0.12.7/bin/node" "/home/jsirois/.nvm/versions/node/v0.12.7/bin/npm" "install" "webpack" "fsevents" "-g" "--prefix" "./webpack-canonical" "--cache" "./npm-cache" "--verbose"
npm ERR! node v0.12.7
npm ERR! npm  v2.11.3
npm ERR! code EBADPLATFORM

npm ERR! notsup Unsupported
npm ERR! notsup Not compatible with your operating system or architecture: fsevents@0.3.8
npm ERR! notsup Valid OS:    darwin
npm ERR! notsup Valid Arch:  any
npm ERR! notsup Actual OS:   linux
npm ERR! notsup Actual Arch: x64
  • e

    enough-analyst-54434

    08/27/2015, 3:44 AM
    I had to search around for a package that had native and built ok on linux, but selenium-webdriver was the ticket. I found this works: ```
  • e

    enough-analyst-54434

    08/27/2015, 3:44 AM
    ```npm install selenium-webdriver@2.46.1 --prefix ./canonical --cache ./npm-cache
  • e

    enough-analyst-54434

    08/27/2015, 3:47 AM
    Um, this:
    Copy code
    npm install selenium-webdriver@2.46.1 --prefix ./canonical --cache ./npm-cache
-
> utf-8-validate@1.1.0 install /tmp/node-test/canonical/node_modules/selenium-webdriver/node_modules/ws/node_modules/utf-8-validate
> node-gyp rebuild

make: Entering directory '/tmp/node-test/canonical/node_modules/selenium-webdriver/node_modules/ws/node_modules/utf-8-validate/build'
  CXX(target) Release/obj.target/validation/src/validation.o
  SOLINK_MODULE(target) Release/obj.target/validation.node
  COPY Release/validation.node
make: Leaving directory '/tmp/node-test/canonical/node_modules/selenium-webdriver/node_modules/ws/node_modules/utf-8-validate/build'

> bufferutil@1.1.0 install /tmp/node-test/canonical/node_modules/selenium-webdriver/node_modules/ws/node_modules/bufferutil
> node-gyp rebuild

make: Entering directory '/tmp/node-test/canonical/node_modules/selenium-webdriver/node_modules/ws/node_modules/bufferutil/build'
  CXX(target) Release/obj.target/bufferutil/src/bufferutil.o
  SOLINK_MODULE(target) Release/obj.target/bufferutil.node
  COPY Release/bufferutil.node
make: Leaving directory '/tmp/node-test/canonical/node_modules/selenium-webdriver/node_modules/ws/node_modules/bufferutil/build'
selenium-webdriver@2.46.1 canonical/node_modules/selenium-webdriver
├── tmp@0.0.24
├── adm-zip@0.4.4
├── xml2js@0.4.4 (sax@0.6.1, xmlbuilder@2.6.4)
├── rimraf@2.4.3 (glob@5.0.14)
└── ws@0.7.2 (options@0.0.6, ultron@1.0.2, utf-8-validate@1.1.0, bufferutil@1.1.0)
jsirois@gill /tmp/node-test $ ls
canonical  npm-cache
jsirois@gill /tmp/node-test $ mkdir copy && cd copy && npm link ../canonical/node_modules/selenium-webdriver
multer@0.1.8 ../canonical/node_modules/selenium-webdriver/node_modules/multer
├── qs@1.2.2
├── mkdirp@0.3.5
├── type-is@1.5.7 (media-typer@0.3.0, mime-types@2.0.14)
└── busboy@0.2.9 (dicer@0.2.3, readable-stream@1.1.13)
... [no SOLINK] ...
promises-aplus-tests@2.1.1 ../canonical/node_modules/selenium-webdriver/node_modules/promises-aplus-tests
├── underscore@1.6.0
├── sinon@1.16.1 (formatio@1.1.1, samsam@1.1.2, lolex@1.3.1, util@0.10.3)
└── mocha@1.21.5 (escape-string-regexp@1.0.2, diff@1.0.8, growl@1.8.1, commander@2.3.0, jade@0.26.3, mkdirp@0.5.0, debug@2.0.0, glob@3.2.3)
/home/jsirois/.nvm/versions/node/v0.12.7/lib/node_modules/selenium-webdriver -> /tmp/node-test/canonical/node_modules/selenium-webdriver
/tmp/node-test/copy/node_modules/selenium-webdriver -> /home/jsirois/.nvm/versions/node/v0.12.7/lib/node_modules/selenium-webdriver -> /tmp/node-test/canonical/node_modules/selenium-webdriver
jsirois@gill /tmp/node-test/copy $ node
> var webdrive = require('selenium-webdriver')
undefined
> webdrive
{ ActionSequence: { [Function] checkModifierKey_: [Function] },
  Builder: [Function],
  By: 
   { className: [Function],
     css: [Function],
     id: [Function],
     linkText: [Function],
     js: [Function],
     name: [Function],
     partialLinkText: [Function],
     tagName: [Function],
     xpath: [Function] },
...
  • e

    enough-analyst-54434

    08/27/2015, 3:48 AM
    So installing once then linking to the path seems to do the trick - the linked-to-thing is not rebuilt
  • w

    witty-crayon-22786

    08/27/2015, 3:48 AM
    perfect
  • w

    witty-crayon-22786

    08/27/2015, 3:49 AM
    (ps: submitqueue timeout due to a pathological hang somewhere in zinc yay. should still go in tomorrow morning though.)
  • e

    enough-analyst-54434

    08/27/2015, 3:49 AM
    Any easy access to jstack the thing?
  • e

    enough-analyst-54434

    08/27/2015, 3:50 AM
    not perfect - I need to dig more - middle bit is not good: 
    /tmp/node-test/copy/node_modules/selenium-webdriver -> /home/jsirois/.nvm/versions/node/v0.12.7/lib/node_modules/selenium-webdriver -> /tmp/node-test/canonical/node_modules/selenium-webdriver
  • e

    enough-analyst-54434

    08/27/2015, 3:50 AM
    the package gets also linked to the global node node_modules
  • w

    witty-crayon-22786

    08/27/2015, 3:51 AM
    probably... but it's not deterministic. need to catch it during the 25 minute window before it gets killed
  • w

    witty-crayon-22786

    08/27/2015, 5:59 PM
    any reason why 
    invalidate_dependents=True
    isn't the default for 
    self.invalidated
    ?
  • e

    enough-analyst-54434

    08/27/2015, 6:09 PM
    It's not what you want to do for some tasks - checkstyle and doc tasks come to mind.. so the direction we default will disfavor one class and favor another either way... suggesting we should force the concious choice and not default the parameters perhaps.
  • e

    enough-analyst-54434

    08/27/2015, 6:10 PM
    The long and short of the issue is dependencies has always been a lie, like fingerprints, dependencies are really a task specific notion
  • w

    witty-crayon-22786

    08/27/2015, 6:11 PM
    yea.
  • w

    witty-crayon-22786

    08/27/2015, 6:25 PM
    @bored-art-40741: do you have time to look at https://rbcommons.com/s/twitter/r/2668/ ?
  • b

    bored-art-40741

    08/27/2015, 6:28 PM
    Looking
  • b

    bored-art-40741

    08/27/2015, 7:20 PM
    Hey @witty-crayon-22786 or @gorgeous-fish-11040 if you're around
  • b

    bored-art-40741

    08/27/2015, 7:20 PM
    I'm having issues with missing dep detection since our recent upgrade
  • b

    bored-art-40741

    08/27/2015, 7:20 PM
    Mostly having to do with how ivy products are spoofed in our custom resolver
  • b

    bored-art-40741

    08/27/2015, 7:21 PM
    But as far as I can tell, we do not have a way to express the notion of a pom-artifact that only exists as a bucket of dependencies
  • b

    bored-art-40741

    08/27/2015, 7:21 PM
    In particular
  • b

    bored-art-40741

    08/27/2015, 7:21 PM
    Copy code
    def get_transitive_jars_by_ref(ref1):
          def create_collection(current_ref):
            return {ivyinfo.modules_by_ref[current_ref].artifact}
          return ivyinfo.traverse_dependency_graph(ref1, create_collection, memo=deps_by_ref_memo)
  • b

    bored-art-40741

    08/27/2015, 7:23 PM
    I think that the collector should be reading from 
    IvyInfo._artifacts_by_ref
    rather than 
    modules_by_ref
1...231232233...2420Latest