beholder icon indicating copy to clipboard operation
beholder copied to clipboard

The `set` constructor for depsets is deprecated and will be removed. Please use the `depset` constructor instead

Open kazemSafari opened this issue 7 years ago • 12 comments

I followed the instructions to install beholder and got to this point: bazel build beholder/tensorboard_x

When i try to run this command i get an error message: ERROR: /home/alleria/beholder/beholder/tensorboard_x/BUILD:34:1: every rule of type ts_web_library implicitly depends upon the target '@com_microsoft_typescript//:tsc', but this target could not be found because of: no such package '@com_microsoft_typescript//': The set constructor for depsets is deprecated and will be removed. Please use the depset constructor instead. You can temporarily enable the deprecated set constructor by passing the flag --incompatible_disallow_set_constructor=false ERROR: Analysis of target '//beholder/tensorboard_x:tensorboard_x' failed; build aborted: no such package '@com_microsoft_typescript//': The set constructor for depsets is deprecated and will be removed. Please use the depset constructor instead. You can temporarily enable the deprecated set constructor by passing the flag --incompatible_disallow_set_constructor=false INFO: Elapsed time: 8.931s FAILED: Build did NOT complete successfully (16 packages loaded)

Is there any fix to this problem? I am using ubuntu 16.04 OS, python3.6, and tensorflow 1.3 CPU version

kazemSafari avatar Nov 15 '17 06:11 kazemSafari

Hi there!

Lame answer: this project is now supported and a part of TensorBoard - you may want to try working through that repo to get it working. Their team now supports this project as a part of TensorBoard. I don't work on it anymore.

Alternative answer: hmmm... have you tried adding that flag (--incompatible_disallow_set_constructor=false) to bazel build?

image https://docs.bazel.build/versions/master/skylark/backward-compatibility.html

chrisranderson avatar Nov 15 '17 06:11 chrisranderson

Hey again chisranderson! Yes! i tried it but it doesn't work and it keeps giving the same error! How could we let tensorboard people know about this issue?!

kazemSafari avatar Nov 16 '17 15:11 kazemSafari

Well, the Beholder they use is different than the one in this repo. They may have a nightly build available with Beholder in it - you can try to see if that is available, install and run it, and see if it has the same problem. After that you could post an issue on the TensorBoard repository.

This repo is the outdated version. :) Maybe I should make that bigger on the README.

chrisranderson avatar Nov 16 '17 15:11 chrisranderson

tb-nightly PyPi isn't available yet, but it will be soon.

jart avatar Nov 16 '17 17:11 jart

Hi again! Thank you for your comments chrisranderson and jart. Well i tried many different methods including installing tf-nightly in a python3.5 virtualenv. All of them failed. I opened a new issue as you suggested in tensorboard repo. I guess i have to wait for tensorboard developers to respond there!
Thanks again!

kazemSafari avatar Nov 18 '17 01:11 kazemSafari

This TensorBoard developer says hello :-)

Part of the issue you describe might be there's no tb-nightly yet to go alongside tf-nightly I'm working on that now. I just got the GCE instance yesterday with which to do it.

jart avatar Nov 18 '17 01:11 jart

Alright! Thank you!

kazemSafari avatar Nov 18 '17 17:11 kazemSafari

@jart Any update on shipping in tf-nightly? I have Beholder working, and tf-nightly working but the dashboard doesn't show. :) Hope things are going well.

chrisranderson avatar Jan 17 '18 18:01 chrisranderson

@jart I'd be fine running tensorboard_x from this repo, but the bazel build fails with all kinds of mismatching checksum errors, or errors about depset that @kazemSafari mentioned.

chrisranderson avatar Jan 17 '18 18:01 chrisranderson

We're in the middle of putting a low latency database in place, so we'll be able to properly maintain Beholder on all systems and environments going forward. We won't be able to devote time to that until early next quarter. Would you be able to help us maintain Beholder until then?

On Jan 17, 2018 10:46 AM, "Chris Anderson" [email protected] wrote:

@jart https://github.com/jart I'd be fine running tensorboard_x from this repo, but the bazel build fails with all kinds of mismatching checksum errors, or errors about depset that @kazemSafari https://github.com/kazemsafari mentioned.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/chrisranderson/beholder/issues/56#issuecomment-358402849, or mute the thread https://github.com/notifications/unsubscribe-auth/AADAbhXqKwDhU7vvpZry9D1iVVL5SsaEks5tLj_2gaJpZM4QecWO .

jart avatar Jan 17 '18 21:01 jart

Gotcha, sounds good.

I actually might not be able to - I'm working full-time now (yay) and it's one of those since-we-give-you-money-we-own-everything-you-build things (boo). I'm in the process of checking on it though.

chrisranderson avatar Jan 17 '18 22:01 chrisranderson

Congratulations on the new gig!

I'd say talk with your manager. You signed the CLA earlier. You'd need to get permission to sign it again under your corp email address.

Not all companies are willing to do this, but the ones that do always end up looking good. It always makes folks happy to see corporations giving back to the open source community. Even if it's only for minor patches. It serves everyone's best interests.

jart avatar Jan 18 '18 01:01 jart