Allow popover content to be lazily instantiated about popover

[julianobrasil]

I'll take a look at them and give it a try. I'll come back to you in case of trouble with the overall process.

[willshowell]

Ah yep, this is not intentional - thanks for the report! I'm afraid it'll have to bump a major release since it seems like a bug folks could accidentally be relying on.

FWIW, if your "content component" is specific enough, you could inject the SatPopover and subscribe to the first opened event before making your ajax request. A workaround still, but potentially more encapsulated.

[willshowell]

@rosslavery I take that back haha. The popover doesn't exist in the DOM until opened, but is instantiate and lives in memory once declared. That's consistent with what I was getting at in #23 (comment).

https://stackblitz.com/edit/sat-popover-issues-qfddyj

Regardless, I agree there should be a lazy way to instantiate it. I'll look into it as time permits!

[rosslavery]

Great thanks. I think material2 just added lazy instantiation for mat-tabs as well as for mat-menu, might wanna take a look at their implementation for inspiration.

[tarekbazine]

... 2 years later. @rosslavery @willshowell any update about this,
I am using the popover in a dynamic built table in which I have all the cells as inline editable, in which I use the popover to show the appropriate form/picker, the problem is that all the popover are loaded ( rows nb * cells nb !!) which makes the table consuming too much resource memory/cpu.

thanks

[julianobrasil]

If you allow the edition of only one cell at a time, you could use just one popover for all the cells and one anchor per cell.

[tarekbazine]

@julianobrasil I tried sharing one popover and then passing the anchor to the cell but the popover was showing but displaying positions was not right, I end up hacking it as suggested in top with *ngIf popover.isOpen() .

[julianobrasil]

I have opened an issue about the wrong position thing a few months ago. Fortunately, there's a workaround for it: #167

[isaackehle]

@julianobrasil There are a number of open issues here and in PRs. I'd be glad to get them resolved if folks can help with the testing.

[julianobrasil]

@pgkehle, do you mean the update of the .spec files?

[julianobrasil]

Or just trying the code of the PRs?

[isaackehle]

No, I mean I tried building locally from scratch some of the different PRs, some things not working. The workaround you mentioned talked about some of the code being merged w/o proper tests, so it seems there's an opportunity to add another PR to fix a couple of these issues. Looking for some help with that.