Re-add file sort by "last date read"

[snelg]

Partial un-do of #10627 : access is not the same as modified, so we want to keep the "last read date" file sort option in addition to the "date modified" option

---

This change is 

[NiLuJe]

See #10627 (comment)

---

I can confirm that behavior, but I'm slightly confused.

Do we touch the file or timestamps ourselves somewhere?

When I tested the atime things the first time we decided to kill it, I was doing synthetic tests w/ stuff like touch, less & vim. And that didn't do much, as expected (noatime & co).

But, if I actually open a book in KOReader (doesn't matter if CRe or PDF), we do affect both its atime and chtime...

(At... slightly different times, as can be seen on a real system with high-res timestamps. And, yes, this is a noatime-mounted tmpfs ;o)).

┌─(niluje@ahsoka:pts/16)───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────(/var/tmp/niluje/koreader/koreader-emulator-x86_64-pc-linux-gnu-debug/koreader)─┐
└─(0.17:18%:03:54:%)── stat /var/tmp/niluje/koreader/test/juliet.epub                                                                                                                                                                                                                                                                                  ──(Wed, Jul 12)─┘
  File: /var/tmp/niluje/koreader/test/juliet.epub
  Size: 839144          Blocks: 1640       IO Block: 4096   regular file
Device: 0,33    Inode: 1262789     Links: 1
Access: (0644/-rw-r--r--)  Uid: ( 1000/  niluje)   Gid: ( 1000/  niluje)
Access: 2023-07-12 03:12:38.000000000 +0200
Modify: 2023-07-12 02:44:23.000000000 +0200
Change: 2023-07-12 03:12:38.533349776 +0200
 Birth: 2023-07-12 02:44:23.825471810 +0200
┌─(niluje@ahsoka:pts/16)───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────(/var/tmp/niluje/koreader/koreader-emulator-x86_64-pc-linux-gnu-debug/koreader)─┐
└─(0.17:18%:03:54:%)── stat /var/tmp/niluje/koreader/test/juliet.epub                                                                                                                                                                                                                                                                                  ──(Wed, Jul 12)─┘
  File: /var/tmp/niluje/koreader/test/juliet.epub
  Size: 839144          Blocks: 1640       IO Block: 4096   regular file
Device: 0,33    Inode: 1262789     Links: 1
Access: (0644/-rw-r--r--)  Uid: ( 1000/  niluje)   Gid: ( 1000/  niluje)
Access: 2023-07-12 03:54:42.000000000 +0200
Modify: 2023-07-12 02:44:23.000000000 +0200
Change: 2023-07-12 03:54:42.496024101 +0200
 Birth: 2023-07-12 02:44:23.825471810 +0200

[NiLuJe]

Do we touch the file or timestamps ourselves somewhere?

Okay, mystery solved ;).

koreader/frontend/readhistory.lua

Line 263 in d57325a

lfs.touch(file, now, mtime) -- update book access time for sorting by last read date

TL;DR: Yeah, this makes sense, and both that and this PR are desirable.

[snelg]

And to be clear, I don't mind if the proper solution is to make it so the mtime updates each time a file is opened, thus making the "date modified" the only needed Date-sorting option.
My patch here was just the easiest code change I saw since it simply un-does a few line changes.

[hius07]

I agree, it makes sense to have both of them.

[NiLuJe]

(Rebased it to make GH happy)