Issue 645207  has been merged into this issue.

This is happening even when I minimize my browser/CrOSH window on lulu 
and even without tweaking any zoom settings. I tried the default as well as higher/lower resolutions-  I hit this problem sooner or later. My work requires to cat large chunks of data from /var/log/messages, so this is a bit annoying, having to hit Ctrl-L after each command. Maybe I should edit my bashrc to append a clear to each command or something? 

Re#3: Press Ctrl+- will show you that last hidden line. 

There seems to be a rounding error in hterm.ScrollPort.prototype.getTopRowIndex() [1]. This function always returns (the correct top row index - 1) when the bug is reproduced. Who owns this code?


[1]: https://cs.corp.google.com/webstore/crx/d/djddabimpcicofmjnlkahgbbpfigocpn/js/nassh_deps.concat.js?type=cs&q=f:+nassh_deps&l=10101

Best to avoid posting links to internal repos in public bugs.

#6: https://chromium.googlesource.com/apps/libapps/+blame/HEAD/hterm/js/hterm_scrollport.js#1178

Last changed here, precisely to try fixing this sort of zoom issue:
https://chromium.googlesource.com/apps/libapps/+/f4e90e86bec06b124eaf4a0c01ea924cdc7f49a4%5E%21/hterm/js/hterm_scrollport.js

wilford/rginda?

I think it just needs to compute the height of character using multiple lines like it does for width.

I fixed it locally as Oshima suggested by calculating the character height. I can send a CL.

hterm/nassh/libapps is all public code.  internal links are irrelevant.

The following revision refers to this bug:
  https://chromium.googlesource.com/apps/libapps/+/7354202ec171287e200ec075c40e34784005fa97

commit 7354202ec171287e200ec075c40e34784005fa97
Author: Ahmed Fakhry <afakhry@google.com>
Date: Wed Sep 14 00:18:48 2016

Fix hterm's last line disappearing depending on the zoom level

We need to average the height of the character the same way we average the width to
avoid rounding errors.

BUG= chromium:633924 
TEST=mkdeps.sh, mkzip.sh --nopromote and then load it on the device and try seq 1000 several times. The last line shouldn't disappear

Change-Id: I0d0dfe749a6e02f8023e2709c7222446d6e136e0
Reviewed-on: https://chromium-review.googlesource.com/385075
Reviewed-by: Rob Ginda <rginda@chromium.org>
Tested-by: Rob Ginda <rginda@chromium.org>

[modify] https://crrev.com/7354202ec171287e200ec075c40e34784005fa97/hterm/js/hterm_scrollport.js

What do we need to do in order to release a new version for Hterm using the fix in #12 to be part of ChromeOS? Or does that happen automatically?

rob or i will take care of it.  we need to do one anyways for the breakage in canary builds.

this seems to be breaking horizontal cursor position tracking

i have a 1599x899 pixel window which is 204x58.  when writing a long line, the cursor slowly falls out of sync.  see attached screenshots -- the cursor should have been to the right of the "f" in every one, but it drifted.

Deleted: cursor-left.png 12.0 KB

	cursor-left.png 12.0 KB View Download

Deleted: cursor-mid.png 12.8 KB

	cursor-mid.png 12.8 KB View Download

Deleted: cursor-right.png 13.0 KB

	cursor-right.png 13.0 KB View Download

that is SecureShell (beta) 0.8.34.3

In order to get this into the default version of crosh that ships as part of the ChromeOS image, you'll have to update crosh.  See https://chromium.googlesource.com/chromiumos/platform/assets/+/master/chromeapps/crosh_builtin/

Re#15: Do you mean the change in #12 broke the cursor tracking?

i'm fairly certain that is the case.  0.8.34.2 works.  the only changes in 0.8.34.3 are:
bf85c3b1f30696dca576bfdcb285ca11055876e5: hterm: disable ligatures
15ed49071379e51ae47d929134366eb156a6943a: copySelectionToClipboard fixes.
7354202ec171287e200ec075c40e34784005fa97: Fix hterm's last line disappearing depending on the zoom level
b976c0b956121d3e10934dea16d61417b6b17f1d: hterm: fix order of parameters to postMessage

seems highly unlikely any of the other commits caused this.

I'll take a look at what happened. I think I have an idea where the problem is.

i've verified that reverting your CL fixes the drift

i'd like to post a new dev version by thur evening, so if you can't fix by then, i'll have to revert.  i'm not saying this as a threat, i just don't have a choice.  there's a bug where SecureShell doesn't work in M54+ and we need to get that fix out by monday which means we need a version in testing before that.  you can always reland.

I've just uploaded a fix.

The following revision refers to this bug:
  https://chromium.googlesource.com/apps/libapps/+/ccbc9ba0f0015da2b545dfb1751496eeeb92a5f6

commit ccbc9ba0f0015da2b545dfb1751496eeeb92a5f6
Author: Ahmed Fakhry <afakhry@google.com>
Date: Thu Sep 15 16:04:24 2016

Fix hterm horizontal cursor position tracking

A regression caused by https://chromium-review.googlesource.com/#/c/385075/3
that resulted in another round error causing the horizontal cursor position
to be wrong when filling the line with characters.
This CL fixes the problem.

BUG= chromium:633924 
TEST=mkdeps.sh, mkzip.sh --nopromote, load it on device, and type
until the line wraps around. Cursor position should remain correct.

Change-Id: Idfc51a1eddf0ef53949822498ac57c89d51d5d7f
Reviewed-on: https://chromium-review.googlesource.com/385916
Reviewed-by: Rob Ginda <rginda@chromium.org>
Tested-by: Rob Ginda <rginda@chromium.org>

[modify] https://crrev.com/ccbc9ba0f0015da2b545dfb1751496eeeb92a5f6/hterm/js/hterm_scrollport.js

I noticed crosh hasn't been updated yet. Should I go ahead and do that?

no need ... i have CLs in flight to rework how crosh is integrated into builds

Awesome. Thanks!

i opened  issue 648422  for that if you want to follow along.  once the CQ starts working again we can start landing things.  then CrOS ToT will follow ToT crosh/hterm automatically.

Verified on Chrome OS 9202.1.0, 57.0.2987.6