webContents.printToPDF rejects in v8.0.0-beta.5

[jacobq]

Preflight Checklist

• I have read the Contributing Guidelines for this project.
• I agree to follow the Code of Conduct that this project adheres to.
• I have searched the issue tracker for an issue that matches the one I want to file, without success.

Note: related but different issues:

• webContents.printToPDF() does not work #21603 -- that one was using the older (callback) API, but this is using the latest (promise) API (and webContents.printToPDF in electron7.x ,it's do not work #21639 appears to be a duplicate of that)
• webContents.printToPDF regression in v5 #18093 -- in that case some output is produced, but here a rejection occurs (and the affected electron versions are different).
• webContents.printToPDF never ends #20634 -- that one never ends, but this one rejects shortly after the call occurs

Issue Details

• Electron Version:
  • v8.0.0-beta.5 (also happens in 9.0.0-nightly.20191220 & 9.0.0-nightly.20200108)
• Operating System:
  • Debian (bullseye/sid) x64, 5.3.0 kernel
• Last Known Working Electron version:
  • 7.1.8, 8.0.0-beta.4, See bisect

Expected Behavior

Calling someBrowserWindow.webContents.printToPDF should not throw exception / generate promise rejection.

Actual Behavior

Calling someBrowserWindow.webContents.printToPDF rejects with [Error: Failed to generate PDF].

To Reproduce

Fiddle: https://gist.github.com/jacobq/cc5e5e1736b88cb32604f55369afd1c7
Repo: https://github.com/jacobq/electron-quick-start/tree/bug/print-to-pdf

mainWindow.webContents.once('did-finish-load', async function() {
  try {
    const pdfData = await mainWindow.webContents.printToPDF({})
    console.log('Got PDF data -->', pdfData) // never gets here
  } catch(e) {
    // output --> Call to printToPDF threw exception [Error: Failed to generate PDF]
    console.error('Call to printToPDF threw exception', e)
  }
});

[jacobq]

FWIW When I bisected, I found 05cde0d to be the first commit with the problem.

Log

git bisect start
# bad: [acd2eabdfd5a5b145b99f336d57ca820b13b48e2] Bump v8.0.0-beta.5
git bisect bad acd2eabdfd5a5b145b99f336d57ca820b13b48e2
# good: [d998bf9feddcfad2889869eb18fdbec7a2832a11] Bump v8.0.0-beta.4
git bisect good d998bf9feddcfad2889869eb18fdbec7a2832a11
# good: [2ab56802cb9ea7549c9020731bdc69618e201f0c] fix: restore POST forms that open a new window with target=_blank (#21474)
git bisect good 2ab56802cb9ea7549c9020731bdc69618e201f0c
# bad: [8dabb3fe335539d46b3a5d635127f073d5d1212b] docs: update installation docs to reflect latest @electron/get changes (#21540)
git bisect bad 8dabb3fe335539d46b3a5d635127f073d5d1212b
# bad: [6c1eb46f685025d34caf07be83d86729e4d67447] Bump v8.0.0-beta.5
git bisect bad 6c1eb46f685025d34caf07be83d86729e4d67447
# good: [696c94c08a5999ac337789883dcfb2028110af7c] fix: sourcemaps not loading with network service (#21493)
git bisect good 696c94c08a5999ac337789883dcfb2028110af7c
# bad: [05cde0d945eb07d388704c34c70878a0f1509d2a] chore: bump chromium to 80.0.3987.6 (8-x-y) (#21061)
git bisect bad 05cde0d945eb07d388704c34c70878a0f1509d2a
# first bad commit: [05cde0d945eb07d388704c34c70878a0f1509d2a] chore: bump chromium to 80.0.3987.6 (8-x-y) (#21061)

[jacobq]

@deepak1556 It looks like #21061 skips the printToPDF test and this is still present in master. Was this an oversight that was supposed to be taken care of before the PR landed? I see that 0c80767 was a "FIXME" commit

[codebytere]

@jacobq i've tracked it down to Chromium CL 1917685 as the culprit; should hopefully have some more progress & answers soon!

[jacobq]

@jacobq i've tracked it down...

Thanks for digging into this. When I discovered that it broke after a Chromium update I threw up my hands and didn't know what to do next (have not yet had to peel back that layer).

[codebytere]

@jacobq PR is up and it should be fixed in the next beta. Chromium is a complicated terror zone so it's not just you 😅