Number of candidates returned for clustering should be capped

[tfmorris]

I have a file where 31,787 clusters were found and the top clusters have 39 variants in them.

We need to cap the number returned to the browser because otherwise it becomes unresponsive (or crashes).

[magdmartin]

So how will you cluster the full group? In several pass so when one click
merge and re-cluster it display the rest of the group?
On 2013-03-06 8:28 PM, "Tom Morris" notifications@github.com wrote:

I have a file where 31,787 clusters were found and the top clusters have
39 variants in them.

We need to cap the number returned to the browser because otherwise it
becomes unresponsive (or crashes).

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/695
.

[tfmorris]

My latest experiment is a 400K row file of menu dishes which are very messy and highly duplicated with the top cluster being 49 ways to spell "fried sweet potatoes." The 41K clusters take 3-7 seconds to compute on the server, but then many minutes to render in my browser.

The default maximum should probably be something like 5-10K, perhaps with a setting to allow it to be increased, like the text facet choices.