So does Cmd-W sometimes not work or does it always work?

I wonder if this isn't "couldn't close", but "was waiting on the page for some reason" (e.g. trying to run an unload handler of some sort) and File>Close tab didn't do that (or by the time it was tried, the page was unblocked, or whatever).

We can't always instantaneously close tabs.  This may be WAI, although perhaps there's a way to give better feedback.

Yeah, sorry for the vague feedback.
It does happen more often when chrome is running really slow.
In these situations Cmd-W doesn't work, but File>Close Tab usually does.
..although, as you were saying, it could be because the page has finished
loading. ..it just usually feels like File>Close tab is somehow doing
something Cmd-W can't.

If this is due to some kind of unload handler, then likely certain pages will tend to exhibit it more frequently.  If you always try cmd-w first and then fall back to the Close Tab menu item, then that could be biasing the perception.  You could try intentionally using Close Tab first.

I don't know the current status of unload handler handling and how it's wired to our different tab-closing options.  Maybe Ojan does.

If it's related to the beforeunload or unload handler, I would expect the tab to close 1 second after you try to close it (or maybe slightly longer, if the machine is under heavy load).  The only time that doesn't work is if you have a DevTools window open for the tab, in which case the timer is disabled.

If that's true, though, I wouldn't expect the X, Command+W, or File->Close Tab to act any differently.

Avi, do you know if File->Close Tab is wired up to a different action than Command+W?  (I would guess that they're the same.)

The traces for the three ways are:

Clicking (X) in tab:

content::WebContentsImpl::~WebContentsImpl() + 14
TabStripModel::InternalCloseTab(content::WebContents*, int, bool) + 313
TabStripModel::InternalCloseTabs(std::__1::vector<int, std::__1::allocator<int> > const&, unsigned int) + 1789
TabStripModel::CloseWebContentsAt(int, unsigned int) + 67
-[TabStripController closeTab:] + 429
-[TabController closeTab:] + 160

File > Close Tab:

content::WebContentsImpl::~WebContentsImpl() + 14
TabStripModel::InternalCloseTab(content::WebContents*, int, bool) + 313
TabStripModel::InternalCloseTabs(std::__1::vector<int, std::__1::allocator<int> > const&, unsigned int) + 1789
chrome::CloseTab(Browser*) + 41
chrome::BrowserCommandController::ExecuteCommandWithDisposition(int, WindowOpenDisposition) + 772
CommandUpdater::ExecuteCommandWithDisposition(int, WindowOpenDisposition) + 312
_os_activity_initiate + 75
-[NSApplication sendAction:to:from:] + 460
-[BrowserCrApplication sendAction:to:from:] + 734
-[NSMenuItem _corePerformAction] + 336
-[NSCarbonMenuImpl performActionWithHighlightingForItemAtIndex:] + 114
_os_activity_initiate + 75
-[NSMenu performActionForItemAtIndex:] + 131
-[NSMenu _internalPerformActionForItemAtIndex:] + 35
-[NSCarbonMenuImpl _carbonCommandProcessEvent:handlerCallRef:] + 107

Command+W:

content::WebContentsImpl::~WebContentsImpl() + 14
TabStripModel::InternalCloseTab(content::WebContents*, int, bool) + 313
TabStripModel::InternalCloseTabs(std::__1::vector<int, std::__1::allocator<int> > const&, unsigned int) + 1789
chrome::CloseTab(Browser*) + 41
chrome::BrowserCommandController::ExecuteCommandWithDisposition(int, WindowOpenDisposition) + 772
CommandUpdater::ExecuteCommandWithDisposition(int, WindowOpenDisposition) + 312
_os_activity_initiate + 75
-[NSApplication sendAction:to:from:] + 460
-[BrowserCrApplication sendAction:to:from:] + 734
-[NSMenuItem _corePerformAction] + 336
-[NSCarbonMenuImpl performActionWithHighlightingForItemAtIndex:] + 114
_os_activity_initiate + 75
-[NSMenu performKeyEquivalent:] + 357
-[NSApplication _handleKeyEquivalent:] + 920
-[NSApplication sendEvent:] + 4274

(X) goes directly to the tab model while the menu (either moused or shortcut keyed) goes through chrome::CloseTab(). These are for about:blank tabs, though, and I don't know if the ones you're using are doing anything fancy.

Thank you for providing more feedback. Adding requester "shrike@chromium.org" for another review and adding "Needs-Review" label for tracking.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Hey sdy@ - can you look into this to see how the Close X path might be failing?