Bugzilla

(In reply to Zhenya Usenko from comment #3)
> Hi. Please look at the fiddle I posted. There are < 20 grid items displayed and we still hit the limit.

Right -- it's not surprising that we hit the limit (and the number of grid items aren't relevant).

The limit here is on row indexes themselves, and has nothing to do with the number of grid items. Your jsfiddle has items with e.g. `grid-row: 9999` through `10010`. Firefox is simply clamping the large *row indexes* so that their ending grid-line is 10,000 (i.e. treating all rows beyond 9999 as 9999).  As noted above, this is allowed by the spec; though we could theoretically relax it (up to a point) if needed.

My question for you (largely out of curiosity) was: what's your use-case for using >10,000 rows (and creating a roughly 280,000px tall grid, in your testcase)?  You mentioned a virtual grid component, but that's still quite-large for a UI component.  Was this something that's extremely-huge to start with, or something like the twitter waterfall that starts off ~screen-sized and gradually grows over time as data gets loaded in, or something else?

(In reply to Zhenya Usenko from comment #4)
> Chrome seems to have a limit of 100000

Do you have a testcase that demonstrates that? I'm seeing different results locally. Though I wonder if maybe there are additional factors beyond raw row count that go into how they limit (if they limit).

Using a modified version of your testcase, I can turn up the row count to over 1,000,000 and they still seem to place the items at distinct positions -- I can load https://jsfiddle.net/uae8sLv2/1 and scroll to the very bottom, and I see 5 distinct bars (at rows 999,998 through 1,000,002).

But 10,000,000 hits some threshold (whether coordinate-space or row-count); I don't see any bars at all at the bottom of https://jsfiddle.net/uae8sLv2/2/ .

(In reply to Zhenya Usenko from comment #3)
> Hi. Please look at the fiddle I posted. There are < 20 grid items displayed and we still hit the limit.

Right -- it's not surprising that we hit the limit (and the number of grid items aren't relevant).

The limit here is on row indexes themselves, and has nothing to do with the number of grid items. Your jsfiddle has items with e.g. `grid-row: 9999` through `10010`. Firefox is simply clamping the large *row indexes* so that their ending grid-line is 10,000 (i.e. treating all rows beyond 9999 as being at the slot between row-line 9999 and row-line 10,000).  As noted above, this clamping is suggested & allowed by the spec; though we could theoretically relax it (up to a point) if needed.

My question for you (largely out of curiosity) was: what's your use-case for using >10,000 rows (and creating a roughly 280,000px tall grid, in your testcase)?  You mentioned a virtual grid component, but that's still quite-large for a UI component.  Was this something that's extremely-huge to start with, or something like the twitter waterfall that starts off ~screen-sized and gradually grows over time as data gets loaded in, or something else?

(In reply to Zhenya Usenko from comment #4)
> Chrome seems to have a limit of 100000

Do you have a testcase that demonstrates that? I'm seeing different results locally. Though I wonder if maybe there are additional factors beyond raw row count that go into how they limit (if they limit).

Using a modified version of your testcase, I can turn up the row count to over 1,000,000 and they still seem to place the items at distinct positions -- I can load https://jsfiddle.net/uae8sLv2/1 and scroll to the very bottom, and I see 5 distinct bars (at rows 999,998 through 1,000,002).

But 10,000,000 hits some threshold (whether coordinate-space or row-count); I don't see any bars at all at the bottom of https://jsfiddle.net/uae8sLv2/2/ .