Show both the course id and the archive file for duplicates when listing course archives in the admin course.

[drgrice1]

The previous setup did not take into account what would happen if multiple archive files contained the same course id. So instead of the listArchivedCourse method returning a hash of the form

{
    myTestCourse => {
        filename => 'myTestCourse.tar.gz',
        size     => '605 KB'
    }
}

it now returns a hash of the form

{
    'myTestCourse.tar.gz' => {
        courseID     => 'myTestCourse',
        size         => '605 KB',
        lastModified => 1778667472
    }
}

Note that the lastModified key is only included because that is what is saved in the cache file, and that is just directly returned if the archive file has not been modified. Although, at this point it is not used elsewhere.

In addition when duplicate courseIDs are seen each of those in the hash will contain the duplicateCourseID key.

Then both the courseID and the filename are then displayed for those archives that are duplicate courseIDs. The format displayed is

notDuplicatedCourseID (2 GB)
myTestCourse (myTestCourse.tar.gz, 605 KB)
myTestCourse (myTestCourseAlt.tar.gz, 1.6 MB)

[drgrice1]

Note that the first commit in this pull request always shows the filename for all archives, and does not have the duplicateCourseID key handling. The second commit adds that. I am not sold on only showing the filename for duplicates, and the second commit could be removed to go back to always showing the archive filenames.

[Alex-Jordan]

I lean toward only printing the file name when there is a difference, for these reasons:

• think that most servers will have only files that match the course ID. So showing the file in addition to the course ID is only adding clutter to most servers.
• When there are mismatches, this would draw attention to the mismatches and help the admin understand which files have a mismatch.

I don't mind if it goes the other way though. If I care later, I can just apply your second commit to my instances.

[somiaj]

I think it is only showing file names of duplicated courseIDs, but I agree it should also show file names if the file name is different than the course ID. Besides for the reasons listed by @Alex-Jordan, it will also help if the admin goes to the file manager looking for a course name1 but it isn't in the list because it was listed under name2.

I don't have strong opinions on if they are always shown (I do like the idea of code simplification) or not, though in most cases if files aren't being renamed or no duplicates, the information is not needed, so I slightly prefer only showing the filename in the case it is different/duplicates exist.

[drgrice1]

So only showing the filename of it differs from the course id is a different way to go. It would be simpler to implement.

[drgrice1]

What is the direction to go?

[Alex-Jordan]

The list of active courses shows course IDs. So my instinct is that the list of archived courses should always show course IDs as well.

And then if that's a course ID that does not match its filename, show the filename. So like:

Courses

1. apple
2. banana

Archived Courses

1. banana (1MB)
2. cherry (2MB)
3. daikon (1MB)
4. daikon (endive.tar.gz 1MB)
5. daikon (fig.tar.gz 2MB)

Again, this is just what I'm thinking. I'm OK with whatever in the end, now that you fixed that slowness issue.

By the way, now that I think about it, why are these ordered lists?

[drgrice1]

The list of archives already and still always shows the course id. These pull requests are about whether or not the archive filename is also shown.

So now we need a vote on this pull request (show both course id and filename for all archive files that share a course id), #3116 (show both course id and filename when the course id and filename are different), or drop the last commit in any of these pull requests and always show both the course id and archive filename.

[Alex-Jordan]

Do I understand right that #3116 (with both commits) would make it like my example earlier? And this one, #3114 with both commits, would look almost the same except it would have 3. daikon (daikon.tar.gz 1MB)?

[Alex-Jordan]

If so, I vote for #3116 with both commits.

[drgrice1]

As to these being ordered lists (I assume you mean on the "Course Listings" page, because on the "Unarchive Course" page they aren't) they have always been ordered lists. Would you rather they were unordered lists?

[somiaj]

This one would also only show cantaloupe (3 MB) even if the file name was different. The other would show cantaloupe (honeydew.tar.gz, 3 MB). I too would rather have #3116. I'll go put my approvals there.

[drgrice1]

With this pull request you would see the filename for all course archives for which another course archive contains the same course id.

With #3116 you see the filename for any course archive for which the filename and course id are different.

[Alex-Jordan]

It just strikes me as odd to enumerate things that are already in an alphabetical order. Having bullets would be worse than numbers for this (imo) though. Let's forget I said anything about that.

[drgrice1]

In any case changing these from an ordered list to something else would be for another pull request and for the next release. I will forget you mentioned it.

[Alex-Jordan]

Closing in favor of #3116.