Ipfs-desktop: placement of dropdown with disappearing task bar is odd

[whyrusleeping]

[hacdias]

Hey @whyrusleeping, this issue should be opened on https://github.com/maxogden/menubar, please. 😄

[machawk1]

@whyrusleeping What OS is this? I don't experience the clipping on macOS 10.13.1 that you showed.

The empirical placement paradigm seems to be that popups are left left with the icon or use some a callout-style triangle to associate an icon with a popup that has a larger width than can be displayed (or simply not left-pinned, e.g., Dropbox).

[whyrusleeping]

Its probably because i set my menubar to autohide.

[hacdias]

Same happens on Windows when the taskbar is set to autohide.

[hacdias]

On Windows, though, when you put the cursor above the application, the taskbar hides again so there's no problem at all. It doesn't happen on macOS too @whyrusleeping ?

[machawk1]

The menu bar where the station icon resides cannot be hidden nor auto-hidden, @hacdias .

[hacdias]

@olizilla please check if this is still an issue with master.

[hacdias]

@olizilla @fsdiogo @machawk1 ping 😄 Could one of you test if this is still happening?

[machawk1]

@hacdias I pulled, compiled, and ran the latest master 529a292bb32886caf9f170d161c11732b8970e3e .

I made sure to hide the menu bar per @whyrusleeping.

The positioning on macOS now seems correct.

[hacdias]

Thanks @machawk1! Closing this then 😄

[machawk1]

A quick followup. After writing the last comment, I relaunched using npm start and experienced some weird positioning behavior. This was tested on a two monitor setup with the left being vertically oriented.

@hacdias I would recommend not closing this issue just yet.

Second run on single horizontal display

Launched on horizontal display, icon selected in menubar on left positioned vertical display.

EDIT: My display configuration diagram per link provided by @hacdias in https://github.com/ipfs-shipyard/ipfs-desktop/issues/529#issuecomment-451198855 :

[hacdias]

Ugh, this is weird. @olizilla @fsdiogo could you take a look into this? It must be an issue of the positioning algorithm or the sizes we get are incorrect.

[hacdias]

Ref.: https://github.com/electron/electron/issues/13461

[hacdias]

Just tried to replicate this and it seems to be solved with #756.