Different markup state for empty/no errors validation summary between server and client implementations

[peterblazejewicz]

I'm not sure why there is a need to render empty list element ('HiddenListItem ') in validation summary tag:

Mvc/src/Microsoft.AspNet.Mvc.ViewFeatures/ViewFeatures/DefaultHtmlGenerator.cs

Line 26 in 538cd9c

private const string HiddenListItem = @"<li style=""display:none""></li>";

Mvc/src/Microsoft.AspNet.Mvc.ViewFeatures/ViewFeatures/DefaultHtmlGenerator.cs

Lines 703 to 707 in 538cd9c

	if (!isHtmlSummaryModified)
	{
	htmlSummary.InnerHtml.AppendEncoded(HiddenListItem);
	htmlSummary.InnerHtml.AppendLine();
	}

When there is no errors the resulting html is:

<div class="text-danger validation-summary-valid" data-valmsg-summary="true">
<ul>
<li style="display:none"></li>
</ul>
</div>

The client side implementation does not care about elements of the list itself, they are always cleared:
https://github.com/aspnet/jquery-validation-unobtrusive/blob/master/jquery.validate.unobtrusive.js#L58-L70

The reason I noticed this is that screen readers/crawlers see that list, even if empty (empty list just adds noise to markup):

It would be better to just not render that empty element on server:

<div class="text-danger validation-summary-valid" data-valmsg-summary="true">
<ul></ul>
</div>

Just a little detail.
Thanks!

[dougbu]

@peterblazejewicz looks like your report is a dupe of #2372. Do you agree?

[peterblazejewicz]

@dougbu
The #2372 seems to be more general discussion about rendering validation summary container, not only its content.
I'm not sure, I just point that rendering empty 'li' is kind of pointless from both server and client side implementation and I think removing it will not have side effects, while clearing rendered markup. Just <li style="display:none;"></li>
Thanks!

[dougbu]

Hmm, the empty <li> element exists to avoid warnings about an empty list (<ul> in this case) and remain compliant with the HTML 4.01 specification.

[peterblazejewicz]

Hmm, the empty <li> element exists to avoid warnings about an empty list (<ul> in this case) and remain compliant with the HTML 4.01 specification.

The data attributes for validation are either not compliant with 4.01 (December 1999)
If older specs are to be supported, then I would opt to close this one and I'll join discussion to #2372.
Thanks!

[dougbu]

The data attributes for validation are either not compliant with 4.01 (December 1999)

I'm curious: In what way are additional attributes non-compliant? The spec doesn't limit the attributes services can include.

I'll join discussion to #2372

You're correct this issue is somewhat separate. Let's leave this open so that I can check w/ others whether the 4.01 spec remains interesting. A few people are out this week.

[peterblazejewicz]

I'm curious: In what way are additional attributes non-compliant? The spec doesn't limit the attributes services can include.

The validation data attributes won't pass W3C 4.01 validators, just like empty UL tags, with 'there is no attribute "DATA-VALMSG-SUMMARY' error (at least when testing document fragments quickly):
https://validator.w3.org/

[Eilon]

Moving to backlog, we will revisit again next time.

[MrJoe]

An unwanted side effect is when using a strict Content-Security-Policy: default-src 'self'; frame-ancestors 'none'; style-src 'self' causes a security violation due to the inline style rendering.

[mkArtakMSFT]

Closing this issue as there was no community involvement for quite a while now.