How can this after-close answer exist?

Question

I recently returned to my computer after a prolonged absence to note a message from the author of this question in response to my comment. While I was there, I noticed the question has been put on hold and had a second answer from the last time I visited. I was going to leave a comment asking the user to not answer questions that were about to be closed for a very clear off-topic reason, when I noticed the answer was posted 1 hour and 52 minutes 18 minutes after the question was put on hold.

Put on Hold: 2017-09-01 18:58:19Z 20:31:46Z

Answer Post: 2017-09-01 20:50:13Z

Is this a bug? What order of operations would make this possible?

Edit: The original version utilized the time at which the question was asked, not the time it was closed. I suspect this is simply a case of the user being in the process of answering the question when it was put on hold, but would like clarification, if possible.

Answer 1

I'm going to blatantly steal Monica Cellio's answer to the same question when it was asked on Mi Yodeya Meta in January 2015, because it puts it far better than I can:

There is a grace period after a question is put on hold during which an answer that was started before the question was put on hold can still be submitted. The rather-old MSE post I linked to says this grace period is around 4 hours, but my own observation has been that it's somewhere around 30-60 minutes nowadays.

Also, apparently the mobile client (I don't know if that's mobile web, apps, or both) doesn't notice that the question was closed while you were answering, so that will always go through.

The linked MSE post is an answer by Tim Stone to How was this answer posted after this question was closed?

So this can happen if someone starts writing an answer before the question gets put on hold, and submits it soon after the question gets put on hold. In this case, the question was put on hold at 20:31:46 UTC and the (reasonably long, even) post-closure answer was posted less than 20 minutes later at 20:50:13 UTC, so given the above, the argument appears to hold.

Adding to the above, How did this user post an answer after the question was put on hold? on MSE has a comment from Oded (another Stack Exchange employee) that says:

This is status-bydesign, so not very likely. The server side check is lenient for a reason. – Oded♦ Jun 28 '15 at 18:42

Answer 2

I’ve seen this discussed on the site-wide meta. The poster can submit the answer late, and it is accepted. Normally the browser communicates with the server and learns that the question closed, and refuses to post, within seconds or a few minutes of it happening. But that's just the browser knowing not to try submitting — if the post was in fact submitted, it would still work, for several hours. This is caused by network glitches and composing while network access is sporatic, etc.