search

mozilla / pdf.js

PDF Reader in JavaScript
https://mozilla.github.io/pdf.js/
Apache License 2.0
47.14k stars 9.82k forks source link

Update jpx decoder #18257

Closed calixteman closed 2 weeks ago

calixteman commented 2 weeks ago

The decoder has been updated to reduce the number of copies after having decoded an image: https://github.com/mozilla/pdf.js.openjpeg/commit/b4ba3b3090347e5efe663f60d61e49f278bfc283

calixteman commented 2 weeks ago

/botio test

moz-tools-bot commented 2 weeks ago

From: Bot.io (Windows)

Received

Command cmd_test from @calixteman received. Current queue size: 0

Live output at: http://54.193.163.58:8877/bb1165482121464/output.txt

moz-tools-bot commented 2 weeks ago

From: Bot.io (Linux m4)

Received

Command cmd_test from @calixteman received. Current queue size: 0

Live output at: http://54.241.84.105:8877/0c11bfa5c26d057/output.txt

moz-tools-bot commented 2 weeks ago

From: Bot.io (Linux m4)

Failed

Full output at http://54.241.84.105:8877/0c11bfa5c26d057/output.txt

Total script time: 28.59 mins

  errors: 1
  different ref/snapshot: 17
  different first/second rendering: 1

Image differences available at: http://54.241.84.105:8877/0c11bfa5c26d057/reftest-analyzer.html#web=eq.log

timvandermeij commented 2 weeks ago

The reference test image difference seem to be limited to the known intermittents, but one error is reported (see below) and that one seems new given that I haven't seen it in earlier runs. Could that be a regression of this patch, or is it a new intermittent?

TEST-UNEXPECTED-FAIL | test failed pr12564 | in chrome | page1 round 1 | render : Error: Requesting object that isn't resolved yet img_p0_15.
calixteman commented 2 weeks ago

The reference test image difference seem to be limited to the known intermittents, but one error is reported (see below) and that one seems new given that I haven't seen it in earlier runs. Could that be a regression of this patch, or is it a new intermittent?

TEST-UNEXPECTED-FAIL | test failed pr12564 | in chrome | page1 round 1 | render : Error: Requesting object that isn't resolved yet img_p0_15.

I don't manage to have an issue while I'm loading this pdf. In considering that there is no jpeg2000 image in this pdf, I'd say it's a new intermittent.

moz-tools-bot commented 2 weeks ago

From: Bot.io (Windows)

Failed

Full output at http://54.193.163.58:8877/bb1165482121464/output.txt

Total script time: 44.30 mins

  different ref/snapshot: 3

Image differences available at: http://54.193.163.58:8877/bb1165482121464/reftest-analyzer.html#web=eq.log

timvandermeij commented 2 weeks ago

I don't manage to have an issue while I'm loading this pdf. In considering that there is no jpeg2000 image in this pdf, I'd say it's a new intermittent.

Thanks for checking! Let's try again in that case and see if it disappears, but given that the Windows bot also didn't fail on this it does look like an intermittent:

/botio-linux browsertest

moz-tools-bot commented 2 weeks ago

From: Bot.io (Linux m4)

Received

Command cmd_browsertest from @timvandermeij received. Current queue size: 0

Live output at: http://54.241.84.105:8877/2f41ad5046e3b05/output.txt

Snuffleupagus commented 2 weeks ago

The reference test image difference seem to be limited to the known intermittents, but one error is reported (see below) and that one seems new given that I haven't seen it in earlier runs. Could that be a regression of this patch, or is it a new intermittent?

TEST-UNEXPECTED-FAIL | test failed pr12564 | in chrome | page1 round 1 | render : Error: Requesting object that isn't resolved yet img_p0_15.

We saw that one fail in PR #18167, where I was a bit worried about the changes there somehow causing it. (Since the image in question should be using a /FlateDecode filter.)

I don't manage to have an issue while I'm loading this pdf.

Well, if this is intermittently failing it might be quite difficult to reproduce (reliably) locally unfortunately...

moz-tools-bot commented 2 weeks ago

From: Bot.io (Linux m4)

Failed

Full output at http://54.241.84.105:8877/2f41ad5046e3b05/output.txt

Total script time: 20.07 mins

  different ref/snapshot: 18
  different first/second rendering: 2

Image differences available at: http://54.241.84.105:8877/2f41ad5046e3b05/reftest-analyzer.html#web=eq.log

timvandermeij commented 2 weeks ago

We saw that one fail in PR https://github.com/mozilla/pdf.js/pull/18167, where I was a bit worried about the changes there somehow causing it. (Since the image in question should be using a /FlateDecode filter.)

That's good to know. Unfortunately then it's apparently so intermittent that we haven't seen it in two weeks time, and I'd also not really have an idea how to reproduce it locally other than perhaps trying to repeatedly open/close the file until it happens.

However, the run above indeed confirmed that it's not caused by this particular PR, so I'll make a follow-up issue for it.

timvandermeij commented 2 weeks ago

Thanks!