BUG: Secondary sorting overrides original sorting

fgh

1) Sort the Library by a column that displays line by line, e g Last Updated.

2) Secondary sort by a column that displays with collapse/expand groups, e g Series.

>> The Library displays as if you had sorted 2) before 1). 

(My Last Updated column was 'cluttered' with so many Texts and Studies: Contributions to Biblical and Patristic Literature volumes that it was hard to see what else was there. I was hoping that a secondary sort by series would keep the date order intact and only sort the series within the dates, thereby conveniently collapsing Texts and Studies into one line.)

I realize this is a bit of a challenge to design and code, but whatever you do with it, it shouldn't behave like this. If you won't make it work, it would be better to refuse the secondary sort and instead display a popup telling us why it can't be done.

Dylan Rondeau

fgh said:

I was hoping that a secondary sort by series would keep the date order intact and only sort the series within the dates, thereby conveniently collapsing Texts and Studies into one line

Unfortunately, this behavior would only make sense when an entire series is updated and has the same date. When that isn't the case (such as making corrections in one volume of a series) you run into this scenario:

• Say we have Series A, with 8 books. We update Book 7 today to correct a typo, but the rest of the series was updated last week.
• We have Series B, with 5 books. The collection was released and downloaded yesterday. 
• If we sorted the way you suggest, it would look like this (with only these 2 series -- imagine this issue with several-thousand-book libraries!):

1. Series A (including 7 books from last week)
2. Series B (5 books from yesterday)
3. This effectively places 7 books out-of-order just to fit the single correctly-sorted item in-order.

Or:

1. A-7 (single)
2. Series B (group)
3. Series A 1-6 (group)
4. A-8 (single)

I can't really see either solution as better than the current implementation, and that's assuming we are even able/willing to mix and match single items with grouped items (which would honestly defeat the purpose of using the grouped sort in the first place).

fgh said:

I realize this is a bit of a challenge to design and code, but whatever you do with it, it shouldn't behave like this. If you won't make it work, it would be better to refuse the secondary sort and instead display a popup telling us why it can't be done.

Honestly I don't think providing no functionality would be better. I believe users (in general) would see one sort direction not being permitted as more of an issue.

MJ. Smith

https://community.logos.com/discussion/comment/609635#Comment_609635

Dylan, your description seems to me to not reflect a secondary sort. In my use of the term, a secondary sort does not change the results of the first. For example if one sorted first on the letter column and did a secondary sort on the numeric column one would get results similar to:

A1
A2
A2
A4
B2
B3
B3
B3
C1
C1
C2
C4

....

or primary sort on number with secondary sort on leter

A1
C1
C1
A2
A2
B2
C2
B3
B3
B3
A4
C4

I agree with fgh that if one cannot preserve the primary sort when doing a secondary sort, the secondary sort should be rejected and the grouping function on the secondary sort should always be dropped.

Dylan Rondeau

https://community.logos.com/discussion/comment/609724#Comment_609724

MJ. Smith said:

Dylan, your description seems to me to not reflect a secondary sort. In my use of the term, a secondary sort does not change the results of the first. For example if one sorted first on the letter column and did a secondary sort on the numeric column one would get results similar to:

...

I agree with fgh that if one cannot preserve the primary sort when doing a secondary sort, the secondary sort should be rejected and the grouping function on the secondary sort should always be dropped.

I think you misunderstood my example. I wasn't doing a sort by letter and number: the number represented book #'s in two series, represented by letter. The complexity comes with the fact that in this case, Series is a sort that, by nature, groups resources. In both of your examples, you split up the items sorted second. The issue here is what to do when that isn't possible (because listing items individually defeats the purpose of sorting by Series at all). I'll restate my example with titles instead. Note that I'm making up the updated dates as an example.

Series: Ancient Egyptian Literature, 3 volumes. Let's say Volume 2 was updated today, and 1 & 3 were updated last week.

Series: The Creeds of Christendom, 3 volumes. Let's say that the entire series was updated yesterday.

Primary sort: Last Updated, Secondary Sort: Title (non-grouping sort)

1. Anc. Egypt Lit: Vol 2   -- today
2. Creeds of Christendom Vol 1   -- yesterday
3. Creeds of Christendom Vol 2   -- yesterday
4. Creeds of Christendom Vol 3   -- yesterday
5. Anc. Egypt Lit: Vol 1   -- last week
6. Anc. Egypt Lit: Vol 3   -- last week

That all makes sense, since we're just ordering individual titles. 

Primary sort: Last Updated, Secondary Sort: Series (problem: grouping sort)

1. Anc. Egypt Lit: Vol 2   -- today
2. Creeds of Christendom: Volumes 1-3   -- yesterday
3. Anc. Egypt Lit: Vol 1   -- last week
4. Anc. Egypt Lit: Vol 3   -- last week

or

1. Anc. Egypt Lit: Volumes 1-3   -- today and last week
2. Creeds of Christendom: Volumes 1-3   -- yesterday

I can see fgh's logic that she would expect Creeds of Christendom to collapse into a group here (since all the dates are the same), but the problem is, how do you handle Anc. Egypt Lit? In order to group them, you either have to leave that series ungrouped (which we don't do, and again defeats the purpose of grouping by series) or violate the primary sort by putting dates out of order to consolidate the series.

I personally doubt that the developers will choose zero functionality over the current implementation, but I can make a case, if you still think that's what you would prefer.

MJ. Smith

https://community.logos.com/discussion/comment/609798#Comment_609798

Dylan Rondeau said:

I think you misunderstood my example.

No, I didn't misunderstand. I was trying to illustrate what one would expect the results of the secondary sort to be. Whether or not the groups are collapsible is a separate question. What I was suggesting was either of two approaches:

1. If one wishes to preserve collapsibility, then one should reject secondary sorts that disrupt it OR
2. If one wishes to preserve a secondary sort function, then one should drop collapsibility when the secondary sort disrupts it

But to try to preserve two logically incompatible functions when the primary sort introduces collapsibillity and the secondary sort disrupts it is setting up expectations that Logos logically can never achieve ... so it is better to be upfront about the problem and which of the alternatives Logos has chosen.

Dylan Rondeau

https://community.logos.com/discussion/comment/609851#Comment_609851

MJ. Smith said:

But to try to preserve two logically incompatible functions when the primary sort introduces collapsibillity and the secondary sort disrupts it is setting up expectations that Logos logically can never achieve ... so it is better to be upfront about the problem and which of the alternatives Logos has chosen.

Okay, I've made the case. Sorry for restating unnecessarily. [:)]