Avoid warning 'calling IsSet with unsupported type "invalid" (<nil>)'

[deining]

This PR addresses #1464.

As pointed out in #608 already, the root cause of the warning is the wrong use of isset.
This PR replaces isset and uses with clauses instead, thus avoiding the confusing warn messages.

This PR closes #1464.

[chalin]

Thanks for looking into this.

There's a simpler way to resolve such expressions, for example:

{{ $sidebarCacheLimit := and .Site.Params.ui .Site.Params.ui.sidebar_cache_limit | default 2000 -}}

It's might be a good idea to also include a test of .Site.Params since that section is absent when, for example, hugo creates a new site.

{{ $sidebarCacheLimit := and .Site.Params .Site.Params.ui .Site.Params.ui.sidebar_cache_limit | default 2000 -}}

Could you change all assignment expressions to be of this form?

[deining]

Thanks for looking into this.

Thanks for reviewing this promptly.

There's a simpler way to resolve such expressions, for example:

{{ $sidebarCacheLimit := and .Site.Params.ui .Site.Params.ui.sidebar_cache_limit | default 2000 -}}

Great idea. I don't see, however, why we additionally should evaluate .Site.Params.ui in this context. .Site.Params.ui is a map, so if .Site.Params.ui.sidebar_cache_limit is set, .Site.Params.ui is set, too. I think we can safely reduce this to:

{{ $sidebarCacheLimit := .Site.Params.ui.sidebar_cache_limit | default 2000 -}}

This is also what the official documentation suggests.

Could you change all assignment expressions to be of this form?

Done.

Meanwhile, I fixed a few unwanted warnings with taxonomies, too.

[chalin]

Have you tested your solution with the params.ui section of the Hugo config commented out?

[deining]

Have you tested your solution with the params.ui section of the Hugo config commented out?

Yes, I did, an it returned the default value, as expected.

[chalin]

Strange. It is as if Hugo's . is acting as a safe selector (like .? in some other languages). I don't think it was always like that, but I'm happy that it works now. Thanks for fixing this.

[chalin]

It might be worth fixing the other occurrences in the use of "isset"?