[Users] Please revert "make Go to/[Next|Prev] sort order aware"

Dave Howorth dave at howorth.org.uk
Mon Mar 27 12:02:44 CEST 2017

On Mon, 27 Mar 2017 11:05:15 +0200
Olivier Brunel <jjk at jjacky.com> wrote:

> Hi all,
> So 3.15.0 was just released, and with it some changes that, while I
> don't know the exact motivations for, I think were meant to make
> things more consistent or met users expectations, but alas turned out
> to be misguided and resulted in opposite behavior.
> Specifically, the idea of making the "Go to Previous/Next" features
> sort order aware. It might sound like a good idea, but it is just
> wrong
> - as in, incorrect.
> Because now the button to go next, with the arrow pointing downwards,
> might actually - and unexpectedly - go up the list. And I can hear you
> about "on purpose" and "sort order", but it's irrelevent.
> When one asks to go to the next unread message, it means two things:
> "unread" says it should be unread, and "next" means it should be next
> on the list, i.e. lower, and that regardless of the current sort
> order.
> The first item on a list is always on top, and the last one at the
> bottom. And to go from first to last, go you down the list by moving
> to the next item, and the next, and so on. How the list is sorted, if
> in fact at all, doesn't matter.

I understand what you're saying but in the context of a date-sorted
list of emails, I expect next-unread in a mail reader to go to a
slightly newer unread email. If I have the view in date-ascending order
that means down, but if I have it in date-descending order (newest at
the top) that means up. I interpret the 'next' in the context of the
mail messages, not how they happen to be displayed.

Now if I'm looking at a threaded view, or one sorted by size (!) I
don't have such strong preconceptions of how mail readers should
behave. And I do agree that a down arrow is not a perfect match to the
semantics, but hey the world isn't perfect. It's just a shape.

Your view is at least as valid as mine, but I just wanted to express at
least one other point of view. It seems to me that there isn't a single
right answer to this question, so perhaps configuration options are the
way to go.

> Imagine you're writing code, and have a list of items. You sort that
> list, by item's date descendingly. Now which item do you expect to
> find first/on top of the list? The more recent one, indeed. And you
> go to the slightly older one by going to the *next* item, of course.
> Doesn't matter what language you're using, what application you're
> using, what the items represent or how sorted the list is, that's
> (always) how it works. The current behavior in claws 3.15.0 makes
> things unexpected, unpredictable, and confusing.
> When enterring a folder and asking to select the *last* item/message,
> it is just wrong - a bug - that claws might, depending on sort order,
> select the *first* item (on top). No, sort order doesn't matter.
> Besides, this also seem to imply an assumption that sort order is
> date-based. What is I sort by size? color label? or whether
> messages have attachments or not? Or does that refer to which
> item/message is "last"? What if the list isn't sorted at all?
> If I sort by size descendingly, I'm specifically asking that the first
> message be the largest one, and the next one be smaller, and the next
> even smaller, and so on. So stop forcing your idea that next equals
> larger, especially when I asked/specified otherwise.
> Let me put it another way: you don't know better than I do what I
> want/need, and that's why you gave me an option - the sort order - so
> I can make/express my choice. Now, please respect it.

The sort order for display is something independent of the order I want
to read messages. They are two separate orderings.
> Thank you.
> _______________________________________________
> Users mailing list
> Users at lists.claws-mail.org
> http://lists.claws-mail.org/cgi-bin/mailman/listinfo/users

More information about the Users mailing list