Mirrors bug mega-compilation (from Slack)

Jeff

Thanks @Jerud for making this list, it does seem urgent. Mirroring would be really useful but it’s just too buggy to use right now.

I just posted a bug report as you were posting this… seems to be similar to the reports from April about “Dragging mirrors doesn’t always put them where they are “dropped””.

Jorge

I think the mirrors functionality is the greatest functionality improvement in the last few months, so spending the time finalizing it to be just right should be a priority.

andrew

Just starting to use mirrors (in order to maintain a list of tasks across multiple projects but where that list has its own sort order) so just discovering issues now.

One I immediately discovered is when I deleted a mirror on mobile, it deleted the original not the mirror, although the mirror still had a 2 next to it and when clicking on it in desktop, it showed the deleted original item in the popup menu.

Jay

Jerud I think it was kind of a chicken and egg problem because it was a fundamentally new organization paradigm. People weren’t ready to change their whole system around an incomplete/buggy feature until it was “ready” so it didn’t get a ton of testing because people were only experimenting with it. Or some (like andrew) just started using mirrors recently.

Additionally the Slack was not as active or diverse as this forum, so there just weren’t that many eyes on it.

Since the Legend launch I’ve now built up a big suite of automated tests and am doing more test driven development, and now as I fix issues I’m adding tests to make sure they don’t come back. So future feature development should be a lot more stable.

But for that chicken and egg problem, maybe we should have two phases of major new feature testing, alpha and beta? So when it’s alpha, it’s something to experiment with but not necessarily ready to really use, and once it’s beta it should be considered safe enough for everyone to use? And that way in the beta phase we’d find more in depth bugs?

Also FYI, these issues have been on my mind too and I’ve been building up a test suite for them. I’m planning to jump into it and try to fix all the mirroring issues this week.

Jay

Thanks for putting this together! The mirror issues have been top of my mind recently and I was just going to dive into a big mirror-fixing project soon, so this is incredibly useful.

christophe

Thank you for this super comprehensive recap.
@Jay will take care of it very quickly, but he promised to solve the problems with the mail panels first !!!!! 😂😂😂😁

andrew

Another for the compilation:

Dragging a mirror onto the calendar creates a new event (seemingly with a link to the mirror?) on the calendar and does not set the date of the mirror or the original item. Deleting that item does delete the new event and the mirror but not the original (no prompt as to what to delete).

Dragging the original to the calendar, as well as setting the date with the date picker of either the original or mirror, does set the date of the original and the mirror.

Jorge

Jerud So I think I was able to reproduce @Jerud ’s July Sometimes Alt dragging creates a link instead of amirror item.

Given a branch A (which has a subtree underneath) that has been mirrored somewhere else (let’s name it branch B) in the outline.
When you position yourself inside the Branch B (the mirror one) and you Alt Drag any sub item within Branch B with the intention to create a mirror inside a mirror, a Link is created instead.
When you position yourself inside the Branch A (the original one) and you Alt Drag any sub item within Branch A with the same intention a mirror is correctly created.

Jerud

Re: Mirrors that appear as Links
I haven’t fully tested all implications of this, but I have observed the following:

Document A contains un-mirrored Parent P with children x, y, z
Document B contains Parent M which is mirrored to another location within Document B
If a child of P is Alt-dragged to nearly anywhere in Document B, it will create a mirror correctly
This includes the original instance of M.
BUT if attempt to create that mirror within the mirrored instance of M, the resulting is drawn as a Link

Now, I can see where you might not like us creating mirrors inside mirrors, but AFAIK this doesn’t create any circular references, and in fact it works just fine for the “original” of M. What’s really messed up about it is that if you take that item and drag it out of P into any other location of Doc B, it will turn back into a Link. And if you try to delete it, you get the popup dialog that says “this item is mirrored…”

Jorge

Another weird behavior with mirrors

Given a branch in document A that contians many sub branches and I want to mirror it to another document.
When I open a new pane on Document B put at the side, filter it for a keyword that matches the parent item where I want to place the mirrored item and then Alt drag the original from Document A to the pane of Document B under the desired parent
Then the original item’s tally shows increased but the mirror item is nowhere to be seen in the destination pane. Only until you remove the filter for the parent, does the mirrored child shows up.
In the screenshot I provided I Alt Dragged the item Medicines from the leftmost pane to the center pane under Project B where it doesn’t appear because it is filtered, however on the 3rd pane (which is pointed to the same document but unfiltered) it appears normally.

this is very disconcerting and it took me a while to figure out what was happening, as it gives the appearance that nothing is happening when you o the alt drag.

Furthermore, when reproducing it in my sandbox to provide a screen shot of an example I noticed that the link to a project I had previously tested shows up in the pane filtered for the keyword but not in the unfiltered pane, which is even weirder. You can see this in my screenshot by the “Project C So this is a link”

Jerud

Jorge that sounds like the main issue is Filters incorrectly excluding Mirrored items, which is a longstanding bug with Mirrors. Your specific use case however highlights exactly how much trouble this can be, as it not only cripples Legend’s functionality but creates the impression of data loss, which is a blow to the app’s reliability.

Jorge

Jorge
So tested the first scenario that I described and it is now working, but my second scenario I mentioned with the link being filtered is still not working.

Jerud

@Jay I’ve been thinking about this thread a lot recently (because I keep getting hosed by mirrors) and it’s got me wondering: How did the “beta” process fail us with this feature? Mirrors went out to beta first of course…but we didn’t get a lot of time with them before they pushed out to stable. I remember it being 3, maybe 4 weeks max? I personally didn’t even try them for at least a week or two after they came out, because I didn’t have time to do that kind of playing. I’m sure that’s true for several others…so it seems to me the feature probably didn’t get much of a trial run at all. Am I off-base with that?

I understand wanting to maintain feature parity with the competition…but this is a big raft of bugs, and every day I worry about new users running into these side effects and getting turned off from the app by them. It could have been headed off by simply…waiting longer. I hope in the future we let new features this substantial hang out in Beta for a bit longer, to really stress-test them properly.

Jerud

Jay Thanks for the background info on your new testing suite and beta process. Alpha–>Beta sounds a bit overwrought to me, I was just thinking leave stuff in Beta for longer. I’m not wary of testing new features (that’s kinda the point of being a beta user) but I can’t always jump right in when they come out. From my perspective, mirrors were not a trivial add or minor tweak, and could have been left to just accumulate usage in beta for a longer period since there was no real rush to get them live. With mirrors, I think I sat on them for a couple weeks because I was just too busy at the time to try them out. Even with Legend, sometimes I get so overwhelmed that I can’t bear the idea of inserting a new wrinkle into my Boards – and I don’t really find bugs in my sandbox. That’s useful to work out reproduction steps and check hunches but I only ever find bugs by trying to fill an actual need in my live Boards.

Re: where to start with Mirrors – my gut response is to fix how they are ignored by filters. That is the most easily encountered issue and is currently breaking my workflows because I set my system up taking for granted that mirrors would be truly treated like any other item. I also presume this would be the most visible – and disconcerting – issue for the broader user base as it can make it look like items are missing or that filters are broken, which are both cornerstones of a users’ trust in the app.

That said, a lot of bugs seem to stem from the existence of an “original” instance. We talked about that before, and I’m not opposed to there being an original “under the hood” but right now it’s not staying under the hood. So perhaps it’s smartest to fix that underlying problem first, since it may change how you address the others? Only you would really know that 😉

Jay

Which of the mirrors issues do you think are the highest priority? Where do you think I should start?

LauraH

Jay My top two mirror bugs:
1 - Zooming goes to the “original”, it should go to the current instance
2 - agree with Jerud - the way they don’t work right in filters is a BIG problem.

LauraH

Another bug for the mirrors compilation: Copy and Paste that involves mirrored items seemed to ignore them.

I’m a bit fuzzy on the specifics; because it took me awhile to realize there was a problem - but I copied and pasted a whole list of items, including several mirrored items, from one document to another.
I realized several days latter - none of the mirrored items pasted at their location. Grrr. Had to go back and hunt down and recreate all the “missing” items.

Jay

I’ve started on mirrors not working right with filters, and it seems like I’ve solved it. But I’m building up a suite of automated tests for filtering to make sure that pane items are always as expected based on the view and filter, so fixing this doesn’t cause other bugs. So I’ll just take a bit of time to keep adding more tests and make sure it’s right before releasing it.