Semantics on number of question marks in a post

Question

My question https://webmasters.stackexchange.com/questions/58048/how-do-cdns-caches-handle-vary-user-agent-header?noredirect=1 was deleted, (not even closed) because a moderator considered it had too many question marks. While I perfectly understand the concept of one question per post, is it unrealistic to ask a question followed by a clarifying question to make sure the original question is not too broad in scope?

Unlike the reason stated by moderator dan, I did not ask for information about "all" CDN providers. I asked for specific ones that people could answer about...And if there were going to be sufficient answers I can see that post being a community wiki answer as a canonical source for this type of information.

Even Matt Cutts briefly covered this topic but from a different angle https://www.youtube.com/watch?v=va6qtaiZRHg

Seems like dan was playing on semantics of pure count of question mark symbols in my question.

I request other moderators of the site to review this decision and reopen my question.

Answer 1

As explained several times, you were asking multiple generalized questions about how CDNs work. Here are the questions you asked, copied verbatim from your question:

How do CDNs' caches handle 'Vary: User-Agent' headers?

Am I correct that with the current configuration proxies/caches won't care about user-agent and will serve whatever they have cached?

However, the negative side-effect of that is that these proxies/caches/CDNs will have to keep separate copies for every single unique user-agent+gzip combination they encounter...and that could be thousands(millions?)

This translates into a lower cache hit ratio, potentially higher load on the origin server and potentially less optimal end user experience. How do various CDNs handle this situation?

Do they truly keep copies for every single combination or do they combine user agents into larger buckets, ie Firefoxes/Chromes/IEs/Operas/Safaris or do they outright ignore this header or is there something in between?

As also explained several times, questions here need to be about a website under your control, not about how all CDNs might work, and suggested that if you edit your question to ask one specific question about one particular CDN in relationship to your site, it would be reviewed by others and possibly be reopened.

Here is my first comment in this regards:

As it stands, this question is asking several generalized questions in regards to CDNs. Please narrow this down to one specific question, about one particular CDN in relationship to your website.

Despite the above, you edited the question to just add bold to one of your questions, and commented back:

@dan this question is asking a single specific question: how CDNs handle Vary: User-Agent header. This question is long because I put in detailed explanation of the situation in which this applies to and for people to understand this issue at hand.

Your question was reviewed again by others and voted to leave it closed.

I also commented back:

You may have added bold to one question, but you're asking several others too. Questions here need to be about a site under your control, not about how all CDNs might work. If you narrow it down to just one question about a particular CDN in relationship to your site, it might be reopened.

You then made another very minor edit, without narrowing down the questions or making it specific to one CDN in relationship to your site.

Again your question was reviewed and voted to leave closed.

You then flagged the question for moderation, complaining that others had voted to close it unfairly, which was declined by another moderator.

Then you left two quite obviously angry comments back-to-back, copied verbatim:

@dan What multiple questions am I asking?! There's only 1, very clear question that I asked in the title and in the body. Does knowing how CDNs handle Vary header is not covered by a scope of knowledge of webmasters in general and this site specifically?!

@dan there! I removed what could possibly be considered "other questions". Now my question has just a single clear question and there's no reason for it to be flagged for closure.

I then left a comment explaining why it was being deleted after all of that:

You're still asking multiple questions (count your question marks) about how all CDNs work, instead of one specific question about a particular CDN in relationship to your site. This was voted closed multiple times, so it's being deleted since you don't seem to be understanding this.

I suggest you refer to the following Help FAQ: Why and how are some questions deleted? As stated there:

Questions that are extremely off topic, or of very low quality, may be removed at the discretion of the community and moderators. Over time, closed questions that are not useful as signpoints to other questions may also be removed, as well as questions which have no significant activity over a very long period after being asked. If you want to improve a question to keep it from being deleted, click the edit button beneath it.

Questions can also be deleted by the community. Moderators can delete any question, and users with sufficient reputation can cast delete votes on closed questions. It takes 3 votes, minimum, to delete a closed question. However, the number of delete votes required scales to the number of votes on the question and all its answers. Questions that have been closed within the past 48 hours cannot be deleted, so as to allow for editing and possible reopening.

Additionally, any answer that accumulates enough offensive or spam flags will be automatically deleted.

Answer 2

I am one of the reviewers of this question and I also decided that it needed to be closed. This question is too broad and does not meet the criteria as a question that should be asked at this website. This website, like the majority of sites in the Stack Exchange network, seeks specific questions citing specific problems. This question, in all of its forms, did not meet that criteria.

The reason why it was deleted is also a valid one. We generally don't delete questions unless they are spam, pure duplicates, or are somehow problematic to leave intact. Your continual editing of the question falls into the latter category. But, the question being deleted really is irrelevant as a closed question cannot receive new answers anyway.

As it stands, this question is not a good fit for this website so its closure is appropriate and the reason to delete it is warranted. I'm sorry but you won't be able to get an answer to your question here.

Answer 3

This is exactly what I'm talking about here! Your commend said:

You're still asking multiple questions (count your question marks) about how all CDNs work, instead of one specific question about a particular CDN in relationship to your site.

You are putting words into my mouth. This is NOT what I asked! As a matter of fact, here's the post the way it is just before you deleted it. (oh ya and do count the question marks in it.

Here it is for the record for others to see:

It is my understanding that HTTP server's Vary header tells intermediate caches/proxies(and CDNs if in use) what parameters of a request can change content of that request. For example our servers return

Vary: Accept-Encoding

that means that page's content can change if browser's gzip support is on or off(ie compressed or uncompressed byte stream). And as such caches/proxies/CDNs will keep 2 different cached versions of that file.

We have pages that serve different content based on User-Agent: header.

With the current configuration proxies/caches won't care about user-agent and will serve whatever they have cached.

The problem with this config is that we could be in a situation where CDNs are serving wrong pages to wrong browsers(for example html5 page to sub-IE8 browsers.)

To resolve this issue we could amend our webserver so that Vary header also contains User-Agent.

Vary: Accept-Encoding,User-Agent

That way caches/proxies/CDNs will keep different versions based on the specific user agent string.

However, the negative side-effect of that is that these proxies/caches/CDNs will have to keep separate copies for every single unique user-agent+gzip combination they encounter...and that could be thousands(millions?)

This translates into a lower cache hit ratio, potentially higher load on the origin server and potentially less optimal end user experience.

How do CDNs handle 'Vary: User-Agent' header?

Do they truly keep copies for every single combination or do they combine user agents into larger buckets, ie Firefoxes/Chromes/IEs/Operas/Safaris or do they outright ignore this header or is there something in between?

Information about specific CDNs implementations would be appreciated.

---

Once again it seems like you are playing on semantics of pure presence of the ? symbol.

But let's go back to the original version and go "question" by "question" to see what they are:

• How do CDNs' caches handle 'Vary: User-Agent' headers?

This is the primary question of my post.

• Am I correct that with the current configuration proxies/caches won't care about user-agent and will serve whatever they have cached?

Originally and removed later on I was confirming my understanding of this very specific subject at hand. Whoever would be answering the above primary question will easily address this one as part of answering the primary one.

• However, the negative side-effect of that is that these proxies/caches/CDNs will have to keep separate copies for every single unique user-agent+gzip combination they encounter...and that could be thousands(millions?)

I see no question asked here. The part with (millions?) was my short way of writing "and that could be thousands or even millions". If you have problems understanding what was written there I question your competency as a moderator. Seems like again you are having a power trip and simply playing on semantics of purely the presence of ? symbol.

• This translates into a lower cache hit ratio, potentially higher load on the origin server and potentially less optimal end user experience. How do various CDNs handle this situation?

Again, I'm simply repeating the original, primary question as part of explaining the background situation of this question.

• Do they truly keep copies for every single combination or do they combine user agents into larger buckets, ie Firefoxes/Chromes/IEs/Operas/Safaris or do they outright ignore this header or is there something in between?

Yet again, it all relates to the original primary question. Here I'm clarifying and narrowing the scope of my question and expected information in the answer.

I'm not asking a hypothetical question. I'm asking a real world question which I am experiencing right now in my use of CDN service from multiple vendors. I'm not asking for a comprehensive list of every single possible CDN vendor. I'm not asking on vendor recommendations. I'm not asking on product suggestions. I'm asking for people to post based on their 1st hand experience or found documentation how any specific CDN provider handles such headers.

---

If you want to play on semantics, lets look your own question on this site https://webmasters.stackexchange.com/questions/49007/pagerank-seo-results-of-using-a-different-domain-for-each-company-product has 3 totally separate questions! To quote yourself "count your question marks". Further more, you are asking about apple.com, iphone.com and imac.com. These are not your sites that you work on and this is not a real problem you are experiencing. You are asking a hypothetical question. Surely it deserves to be DELETED...let's see what happens.