Given that it's an action normally reserved for useless comments or threads, whilst not being as extreme as delete, shouldn't it have a Yellow scrollover? Right now it's just the normal gray.
I love the look of the Preview now, much better than the old yellow box! I'd think about adding an edit icon instead of the current "Keep editing" and the current Tipsy notification. Or maybe just clicking on the post to edit it. The current implementation seems overly complicated.
Nicereddy: We wanted to make sure that no one ever felt that it was a thing - that there was a warning, or that it could be a negative action. New users, random folks, they should be able to hide things without the psychological weight.
When you talk about deleting and suppressing, however, we want to make sure that people thing twice about it before it happens. Hence the coloration of those events.
Jorm (WMF): But you feel that editing comments/thread titles is a "negative action"? In fact, it is colored orange, which is more severe than yellow. Also, close topic is colored red, which implies a highly negative action. While closing and editing shouldn't be considered "negative", hiding really should.
This box says Reply to "Shouldn't "Hide topic" be yellow?" I thought that had been sorted already.
Okay....so it's ignoring the markup so that I can actually display what I see in the "reply to" box below, but it instead of an apostrophe I have ' then a semi-colon and instead of quotation marks I have " and semi-colon.
Again, I thought that had been sorted...months ago.
Risker: There was a front-end rewrite, which is this specific release. Bugs and regressions creep back in; that's why they're called "regressions".
Jorm (WMF): Well, yes. On the other hand, this is the sort of thing that should show up on testing before the release is made accessible. "Problems we already solved" should be tested to make sure that they're still solved.
(Just a side note here to indicate that, despite having "email me if a page on my watchlist is edited" and having all of the email notification options checked, I didn't receive any emails from Mediawiki in the last 24 hours, despite multiple responses on this page. Not sure if it's me or something else.)
Risker: mediawiki is pretty much the "test bed". I thought that was fairly clear; just banging around on our other wikis isn't going to give us enough range of situation.
Jorm (WMF): Perhaps true, however regression testing is a pretty important part of pre-release testing, and it does not appear to be happening in any organized way; throwing something up on Mediawikiwiki isn't particularly effective if nobody tests for previously resolved problems. It's a topic I find quite interesting, but I don't have the background knowledge or hours in the day to really research and propose solutions. I just know from a user perspective that when issues that seemed to have been resolved long before come back, it's really frustrating and makes me wonder what else is wrong that I haven't spotted.
Risker: There's probably lots you haven't spotted. This is how software development works. This version has been deployed here for less than a week; that's not enough time to discover everything. And the bug you're worried about (a failed triple-level ascii encoding) wouldn't show up in automated tests.
You seem to be taking this with much more emotion than is needed. Nothing is perfect on day one, nor on day fifty.
Jorm (WMF): Nah, I'm not particularly emotional about it. I'm more curious about it, I really don't understand how the WMF is doing regression testing. I sometimes do regression testing on some proprietary software I use for work, but the test instructions are very detailed and often include steps that to the unimaginative aren't obviously connected with the upgrade/patch. Failed regression testing is a rather big deal in some commercial software, and will often result in delays to releases/upgrades, if for no other reason than to avoid very expensive rollbacks. I'm not going to kid myself that a regression in MediaWiki is ever going to have the same impact.
