PageController to write out Last-Modified response header about dxa-web-application-dotnet HOT 4 OPEN

TSI-1943 (internal issue ID for tracking purposes)

from dxa-web-application-dotnet.

We have been looking into this Enhancement further (we were considering implementing it in DXA 1.7), but found it to be rather tricky:
Obtaining the Page's last publish date from Broker/DD4T and returning it as Last-Modified header is quite easy, but if the Page Model is dynamically modified (e.g. by XO Model Builder, by CX Dynamic Entity Evaluator or by ListController) or contains DCPs, using the underlying Page's last publish date may not be appropriate.

Closely related to the Last-Modified response header is processing of the If-Modified-Since request header. For this, we have to be especially careful with dynamic Pages: how do we reliably determine whether a rendered Page is modified if it contains dynamic data (which may be obtained from other systems) and even the Views may include dynamic data?

Suggestions how to deal with this?

from dxa-web-application-dotnet.

This functionality would be used (in our case at least) in combination with a deployer extension that flushes the URL on page publish. In this use case using the publish date as Last-Modified header would be perfectly acceptable. Also pages with dynamic content should probably not be cached by the CDN to begin with :).

IMHO I think trying to determine a Last-Modified date depending on when dynamic content on a page changes would be sort of a theoretical discussion and sounds to me like over-engineering the solution. I would compare it to the lastmod time stamp of a page in the sitemap.xml; I don't think anyone generates those real time based on dynamic content on a page.

from dxa-web-application-dotnet.

The question is whether it is useful to put a (potentially wrong) Last-Modified response header in, without processing If-Modified-Since request headers?
Typically, the Last-Modified response header is used to trigger a If-Modified-Since header on subsequent requests and the server can then send an HTTP 304 Not Modified response if the page wasn't modified since that date/time.
However, accurately determining whether a (potentially dynamic) Page is modified is not easy.

I agree that "pages with dynamic content should probably not be cached (by the CDN) to begin with". But how do we distinguish between "pages with dynamic content" and "pages with static content"? Isn't the whole idea behind Dynamic Delivery architectures like DXA that the majority of pages have (more or less) dynamic content?

from dxa-web-application-dotnet.