New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

No edges appear in the graph explorer for Chrome 62. #425

Closed
chihuahua opened this Issue Aug 26, 2017 · 5 comments

Comments

Projects
None yet
3 participants
@chihuahua
Member

chihuahua commented Aug 26, 2017

bnojao5eudk

The problem is that Chrome 62 no longer supports the native getPathSegAtLength method. SVG 2 dropped the method. We should instead make use of getTotalLength() and getPointAtLength() on the SVGGeometryElement object.

This issue must be fixed quickly, and then we should release a new pypi package as well as internal version. Otherwise, once Chrome 62 enters stable, no data flow edges will appear in the graph explorer for users (the graph would be very much broken). Chrome 62 has already entered the dev channel.

@chihuahua chihuahua self-assigned this Aug 26, 2017

chihuahua added a commit that referenced this issue Aug 27, 2017

Implement getPathSegAtLength in JavaScript
In SVG 2, getPathSegAtLength (which the graph explorer depends on as part of
rendering edges), is deprecated. This means that the graph explorer
would render with no edges in Chrome 62 (current in the dev channel) and
would thus be very much broken in Chrome 62.

This commit pre-empts that change by implementing the functionality of
getPathSegAtLength in the graph explorer so that the explorer still
works in Chrome 62.

I do not see alternative ways of implementing this functionality using
the native pathLength property + getTotalLength() and getPointAtLength()
methods still provided by the SVG API.

After this change goes in, we should quickly release a new pypi package
and internal version of TensorBoard so that the graph explorer is not
broken for users of Chrome 62.

Fixes #425.

chihuahua added a commit that referenced this issue Aug 30, 2017

Implement getPathSegAtLength in JavaScript (#434)
In SVG 2, getPathSegAtLength (which the graph explorer depends on as part of
rendering edges), is deprecated. This means that the graph explorer
would render with no edges in Chrome 62 (current in the dev channel) and
would thus be very much broken in Chrome 62.

This commit pre-empts that change by implementing the functionality of
getPathSegAtLength in the graph explorer so that the explorer still
works in Chrome 62.

I do not see alternative ways of implementing this functionality using
the native pathLength property + getTotalLength() and getPointAtLength()
methods still provided by the SVG API.

After this change goes in, we should quickly release a new pypi package
and internal version of TensorBoard so that the graph explorer is not
broken for users of Chrome 62.

Fixes #425.

jart added a commit to jart/tensorboard that referenced this issue Sep 6, 2017

Implement getPathSegAtLength in JavaScript (tensorflow#434)
In SVG 2, getPathSegAtLength (which the graph explorer depends on as part of
rendering edges), is deprecated. This means that the graph explorer
would render with no edges in Chrome 62 (current in the dev channel) and
would thus be very much broken in Chrome 62.

This commit pre-empts that change by implementing the functionality of
getPathSegAtLength in the graph explorer so that the explorer still
works in Chrome 62.

I do not see alternative ways of implementing this functionality using
the native pathLength property + getTotalLength() and getPointAtLength()
methods still provided by the SVG API.

After this change goes in, we should quickly release a new pypi package
and internal version of TensorBoard so that the graph explorer is not
broken for users of Chrome 62.

Fixes tensorflow#425.

jart added a commit that referenced this issue Sep 6, 2017

Implement getPathSegAtLength in JavaScript (#434)
In SVG 2, getPathSegAtLength (which the graph explorer depends on as part of
rendering edges), is deprecated. This means that the graph explorer
would render with no edges in Chrome 62 (current in the dev channel) and
would thus be very much broken in Chrome 62.

This commit pre-empts that change by implementing the functionality of
getPathSegAtLength in the graph explorer so that the explorer still
works in Chrome 62.

I do not see alternative ways of implementing this functionality using
the native pathLength property + getTotalLength() and getPointAtLength()
methods still provided by the SVG API.

After this change goes in, we should quickly release a new pypi package
and internal version of TensorBoard so that the graph explorer is not
broken for users of Chrome 62.

Fixes #425.
@lipond

This comment has been minimized.

Show comment
Hide comment
@lipond

lipond Jan 10, 2018

so what's the solution?

lipond commented Jan 10, 2018

so what's the solution?

@chihuahua

This comment has been minimized.

Show comment
Hide comment
@chihuahua

chihuahua Jan 10, 2018

Member

I fixed this problem in #434! Upgrade to the latest tensorboard. Thanks!

Member

chihuahua commented Jan 10, 2018

I fixed this problem in #434! Upgrade to the latest tensorboard. Thanks!

@lipond

This comment has been minimized.

Show comment
Hide comment
@lipond

lipond Jan 11, 2018

hi @chihuahua, is there any way to upgrade tensorboard while not upgrading tensorflow?

lipond commented Jan 11, 2018

hi @chihuahua, is there any way to upgrade tensorboard while not upgrading tensorflow?

@jart

This comment has been minimized.

Show comment
Hide comment
@jart

jart Jan 11, 2018

Member

It might work if you pip install tb-nightly

Member

jart commented Jan 11, 2018

It might work if you pip install tb-nightly

@chihuahua

This comment has been minimized.

Show comment
Hide comment
@chihuahua

chihuahua Jan 11, 2018

Member

Ah yeah! Per @jart, tb-nightly gives you bleeding edge features of TensorBoard.

Member

chihuahua commented Jan 11, 2018

Ah yeah! Per @jart, tb-nightly gives you bleeding edge features of TensorBoard.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment