History State API #15261

cbrewster · 2017-01-27T05:31:23Z

This PR implements the state-related portions of the History API. We need to think out how exactly to handle the history state in the case of 1. Reloads and 2. Pruning/Reloading Distant History.

./mach build -d does not report any errors
./mach test-tidy does not report any errors
These changes fix #__ (github issue number if applicable).

There are tests for these changes OR
These changes do not require tests because _____

This change is

highfive · 2017-01-27T05:31:27Z

Heads up! This PR modifies the following files:

@asajeffrey: components/constellation/frame.rs, components/constellation/constellation.rs
@fitzgen: components/script_traits/lib.rs, components/script_traits/lib.rs, components/script/dom/popstateevent.rs, components/script_traits/script_msg.rs, components/script_traits/script_msg.rs, components/script/dom/history.rs, components/script/script_thread.rs, components/script/dom/webidls/History.webidl, components/script/dom/bindings/trace.rs
@KiChjang: components/script_traits/lib.rs, components/script_traits/lib.rs, components/script/dom/popstateevent.rs, components/script_traits/script_msg.rs, components/script_traits/script_msg.rs, components/script/dom/history.rs, components/script/script_thread.rs, components/script/dom/webidls/History.webidl, components/script/dom/bindings/trace.rs

highfive · 2017-01-27T05:31:28Z

Warning

These commits modify unsafe code. Please review it carefully!

cbrewster · 2017-01-27T05:31:38Z

r? @asajeffrey

cbrewster · 2017-01-27T05:32:55Z

@bors-servo try

bors-servo · 2017-01-27T05:32:59Z

⌛ Trying commit c9582e7 with merge 4b08440...


        Auto merge of #15261 - cbrewster:finish_the_history, r=<try>

History State API  This PR implements the state-related portions of the History API. We need to think out how exactly to handle the history state in the case of 1. Reloads and 2. Pruning/Reloading Distant History. ---  - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [ ] These changes fix #__ (github issue number if applicable).  - [X] There are tests for these changes OR - [ ] These changes do not require tests because _____   --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/15261)

cbrewster · 2017-01-27T05:38:22Z

...rsal/persisted-user-state-restoration/scroll-restoration-fragment-scrolling-samedoc.html.ini

  [Manual scroll restoration should take precedent over scrolling to fragment in cross doc navigation]
-    expected: FAIL
+    expected: TIMEOUT


This new TIMEOUT is due to hashchange events not being implemented yet.

cbrewster · 2017-01-27T05:38:22Z

...ry-traversal/persisted-user-state-restoration/scroll-restoration-navigation-samedoc.html.ini

  [history.{push,replace}State retain scroll restoration mode and navigation in the same document respects it]
-    expected: FAIL
+    expected: TIMEOUT


Same as above

bors-servo · 2017-01-27T06:20:53Z

💔 Test failed - linux-rel-wpt

cbrewster · 2017-01-27T06:22:32Z

  ▶ Unexpected subtest result in /html/dom/interfaces.html:
  └ PASS [expected FAIL] History interface: attribute state

  ▶ Unexpected subtest result in /html/dom/interfaces.html:
  └ PASS [expected FAIL] History interface: operation pushState(any,DOMString,DOMString)

  ▶ Unexpected subtest result in /html/dom/interfaces.html:
  └ PASS [expected FAIL] History interface: operation replaceState(any,DOMString,DOMString)

  ▶ Unexpected subtest result in /html/dom/interfaces.html:
  └ PASS [expected FAIL] History interface: window.history must inherit property "state" with the proper type (2)

  ▶ Unexpected subtest result in /html/dom/interfaces.html:
  └ PASS [expected FAIL] History interface: window.history must inherit property "pushState" with the proper type (6)

  ▶ Unexpected subtest result in /html/dom/interfaces.html:
  └ PASS [expected FAIL] History interface: calling pushState(any,DOMString,DOMString) on window.history with too few arguments must throw TypeError

  ▶ Unexpected subtest result in /html/dom/interfaces.html:
  └ PASS [expected FAIL] History interface: window.history must inherit property "replaceState" with the proper type (7)

  ▶ Unexpected subtest result in /html/dom/interfaces.html:
  └ PASS [expected FAIL] History interface: calling replaceState(any,DOMString,DOMString) on window.history with too few arguments must throw TypeError

cbrewster · 2017-01-27T16:37:01Z

@bors-servo try

bors-servo · 2017-01-27T16:37:07Z

⌛ Trying commit 979e719 with merge 886c7a7...


        Auto merge of #15261 - cbrewster:finish_the_history, r=<try>

History State API  This PR implements the state-related portions of the History API. We need to think out how exactly to handle the history state in the case of 1. Reloads and 2. Pruning/Reloading Distant History. ---  - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [ ] These changes fix #__ (github issue number if applicable).  - [X] There are tests for these changes OR - [ ] These changes do not require tests because _____   --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/15261)

bors-servo · 2017-01-27T17:14:20Z

💔 Test failed - linux-rel-wpt

bors-servo · 2017-01-28T01:28:34Z

☔ The latest upstream changes (presumably #14971) made this pull request unmergeable. Please resolve the merge conflicts.

cbrewster · 2017-01-29T05:32:29Z

@jdm I have been looking at how to handle the StructuredCloneData. I have tried using an enum that stores either StructuredCloneData or a cached Heap<JSVal>. This enum is stored in a HistoryEntry which is stored in a HashMap on History. The issue is that every time I read from StructuredCloneData, it always gives me a garbage value. I noticed that Gecko is using a StructuredCloneContainer, I am not sure if that is just a wrapper over StructuredCloneData or if its doing something more complicated.

cbrewster · 2017-02-10T02:33:26Z

@bors-servo try

bors-servo · 2017-02-10T02:33:29Z

⌛ Trying commit 677f30c with merge 8695eb1...


        Auto merge of #15261 - cbrewster:finish_the_history, r=<try>

History State API  This PR implements the state-related portions of the History API. We need to think out how exactly to handle the history state in the case of 1. Reloads and 2. Pruning/Reloading Distant History. ---  - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [ ] These changes fix #__ (github issue number if applicable).  - [X] There are tests for these changes OR - [ ] These changes do not require tests because _____   --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/15261)

bors-servo · 2017-02-10T03:34:18Z

💔 Test failed - linux-rel-wpt

cbrewster · 2017-02-13T03:54:13Z

Reusing the same pipeline id does cause some other unintended issues as Pipelines are not closed right away. They are closed after the constellation received the pipeline exited message from the script thread. Having an entry keeping the pipeline id causes some unintended messages to be send. There are also cases where we may discard a pipeline and traverse back to that entry that had that pipeline before the pipeline was completely closed.

After thinking about it, I wonder if we could something along these lines:

Use a Rc<Cell<Option<PipelineId>>> to group entries that share pipelines. This way when we discard the pipeline of an entry, we can set the Option to None and that will in turn update all the entries that share the same document. The only issue with this is we currently store a PipelineId on Frame which would break the reference.
This will be a bit invasive when dealing with updating the pipeline with the newly created pipeline. We currently use the same logic when navigating with replacement enabled and reloading the document that multiple entries share. Navigating with replacement should create a new pipeline and only affect the single reloaded entry. Reloading a discarded Pipeline should update all entries that previously shared the pipeline.

cbrewster · 2017-02-15T16:49:28Z

Blocked on #15566

asajeffrey · 2017-02-16T01:08:29Z

Sounds like recycling pipeline IDs is having unexpected consequences.

bors-servo · 2017-02-16T12:49:44Z

☔ The latest upstream changes (presumably #15537) made this pull request unmergeable. Please resolve the merge conflicts.


        Link frame entries together that share pipelines

Various frame-related cleanup


        Add constellation frame unit tests


        Implement history states and hash changes.

This change implements handling history state entries and hash change history entries. The constellation is in charge of managing session history and traversals. The History object is in charge of updating all state on the script/dom side of things. The resource thread stores the serialized history state data. This data must be stored outside the script thread as the script thread may be shut down if its document is purged during history pruning. Storing the state data allows for retrieval of the state data at a later time even if the document had to be recreated. Each state is tracked using a state ID which is a UUID. Traversal tasks are now queued on the history traversal task source. The runnable waits synchronously on the constellation to finish handling the traversal. This allows for tests to use setTimeout to wait for a traversal to be handled before the callback has a chance of being called.


        Enable history interface 001 and 002 tests


        Add extra time for document url to be updated


        Update test expectations

update test expectations


        Add frame state id unit test

bors-servo · 2017-02-18T21:28:03Z

☔ The latest upstream changes (presumably #15588) made this pull request unmergeable. Please resolve the merge conflicts.

asajeffrey · 2017-02-24T02:38:45Z

IRC conversation about session histories with multiple instances of the same pipeline id: http://logs.glob.uno/?c=mozilla%23servo&s=24+Feb+2017&e=24+Feb+2017#c619161

TL;DR: If you have document A which pushStates to B then again to C, then their shared Document is discarded, then the history is traversed to B, what happens? Our read of the spec is that A, B and C should all be reloaded to the same fresh document. This means that any session history entries with the same pipeline id must be contiguous.

(Duplicating #10992 (comment) just to make sure I don't forget.)

nox · 2017-04-04T11:39:05Z

What is the status on this? If this is stalled, could we close the PR and file an issue referencing it?

cbrewster · 2017-04-04T13:43:51Z

@nox right now we need to find a better way of handling distant history pipeline discard that allows the history to contain multiple entries with the same pipeline id. I'm finishing up school at the moment so I don't have time right now, but I plan on finishing this up here in a couple weeks.

I'll close for now until I have time to resume work.

highfive assigned wafflespeanut Jan 27, 2017

highfive added the S-awaiting-review label Jan 27, 2017

highfive assigned asajeffrey and unassigned wafflespeanut Jan 27, 2017

cbrewster mentioned this pull request Jan 27, 2017

[WIP] Implement History State #13781

Closed

3 of 5 tasks complete

cbrewster reviewed Jan 27, 2017

View changes

cbrewster mentioned this pull request Jan 27, 2017

History API metabug #10992

Open

18 of 24 tasks complete

highfive added the S-tests-failed label Jan 27, 2017

cbrewster force-pushed the cbrewster:finish_the_history branch from c9582e7 to ff16b67 Jan 27, 2017

highfive removed the S-tests-failed label Jan 27, 2017

cbrewster force-pushed the cbrewster:finish_the_history branch 3 times, most recently from bb10d3d to 979e719 Jan 27, 2017

highfive added the S-tests-failed label Jan 27, 2017

highfive added the S-needs-rebase label Jan 28, 2017

cbrewster force-pushed the cbrewster:finish_the_history branch from 979e719 to 30a1d59 Feb 1, 2017

highfive added the S-tests-failed label Feb 10, 2017

cbrewster force-pushed the cbrewster:finish_the_history branch from 677f30c to 4e2f6af Feb 10, 2017

highfive removed the S-tests-failed label Feb 10, 2017

cbrewster added S-blocked-on-external and removed S-awaiting-review labels Feb 15, 2017

highfive added the S-needs-rebase label Feb 16, 2017

cbrewster added 7 commits Feb 15, 2017

Link frame entries together that share pipelines

326933c

Various frame-related cleanup

Add constellation frame unit tests

Loading status checks…

e64556b

Enable history interface 001 and 002 tests

8d4d854

Add extra time for document url to be updated

79e39b0

Update test expectations

ce76ac0

update test expectations

Add frame state id unit test

Loading status checks…

13a9509

cbrewster force-pushed the cbrewster:finish_the_history branch from 4e2f6af to 13a9509 Feb 17, 2017

highfive added the S-awaiting-review label Feb 17, 2017

cbrewster removed the S-needs-rebase label Feb 17, 2017

highfive added the S-needs-rebase label Feb 18, 2017

cbrewster closed this Apr 4, 2017

servo / servo

History State API #15261

History State API #15261

cbrewster commented Jan 27, 2017 •

edited by larsbergstrom

highfive commented Jan 27, 2017

highfive commented Jan 27, 2017

cbrewster commented Jan 27, 2017

cbrewster commented Jan 27, 2017

bors-servo commented Jan 27, 2017

This comment has been minimized.

This comment has been minimized.

bors-servo commented Jan 27, 2017

cbrewster commented Jan 27, 2017

cbrewster commented Jan 27, 2017

bors-servo commented Jan 27, 2017

bors-servo commented Jan 27, 2017

bors-servo commented Jan 28, 2017

cbrewster commented Jan 29, 2017

cbrewster commented Feb 10, 2017

bors-servo commented Feb 10, 2017

bors-servo commented Feb 10, 2017

cbrewster commented Feb 13, 2017 •

edited

cbrewster commented Feb 15, 2017

asajeffrey commented Feb 16, 2017

bors-servo commented Feb 16, 2017

bors-servo commented Feb 18, 2017

asajeffrey commented Feb 24, 2017

nox commented Apr 4, 2017

cbrewster commented Apr 4, 2017

servo / servo

Join GitHub today

History State API #15261

History State API #15261

Conversation

cbrewster commented Jan 27, 2017 • edited by larsbergstrom

highfive commented Jan 27, 2017

highfive commented Jan 27, 2017

cbrewster commented Jan 27, 2017

cbrewster commented Jan 27, 2017

bors-servo commented Jan 27, 2017

This comment has been minimized.

cbrewster Jan 27, 2017

This comment has been minimized.

cbrewster Jan 27, 2017

bors-servo commented Jan 27, 2017

cbrewster commented Jan 27, 2017

cbrewster commented Jan 27, 2017

bors-servo commented Jan 27, 2017

bors-servo commented Jan 27, 2017

bors-servo commented Jan 28, 2017

cbrewster commented Jan 29, 2017

cbrewster commented Feb 10, 2017

bors-servo commented Feb 10, 2017

bors-servo commented Feb 10, 2017

cbrewster commented Feb 13, 2017 • edited

cbrewster commented Feb 15, 2017

asajeffrey commented Feb 16, 2017

bors-servo commented Feb 16, 2017

bors-servo commented Feb 18, 2017

asajeffrey commented Feb 24, 2017

nox commented Apr 4, 2017

cbrewster commented Apr 4, 2017

cbrewster commented Jan 27, 2017 •

edited by larsbergstrom

cbrewster commented Feb 13, 2017 •

edited