0

I think the tool that allows git to display huge chunks of text in pages on the console is called the "pager". With it, one can scroll up and down using the arrow keys or the page keys, and even sideways to see text that extends beyond the right side of the console.

This had been working great for me until git 2.32.

Now, scrolling down with the "down" arrow key behaves very strangely. New lines at the bottom are limited to some magical maximum width, so that if the next line of text is longer than this width, a portion of it the size of the magical width is displayed on the same console line with every press of the "down" arrow, until the line content is exhausted. What is therefore left is the last chunk of the source text, aligned to the left.

Then a new console line is created, and the first chunk of the next line of text is displayed there. Always with a colon at the end as well (to signify that more of the line will be shown next ?)

 Mon Jun 28 14:41:50 2021) <***author***>
backs added to data (9 days ago, Mon Jun 28 11:12:34 2021) <***author***>
days ago, Mon Jun 28 09:06:41 2021) <***author***>
22:56 2021) <***author***>
minor fixes (2 weeks ago, Wed Jun 23 20:58:59 2021) <***author***>
:00 2021) <***author***>
| | | | | | | | | | | | | | | | | | | | |\ \ \ \ \
| | | | | |_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|/ / / / /
| | | | |/| | | | | | | | | | | | | | | | | | | |
r fixes (3 weeks ago, Thu Jun 17 19:26:42 2021) <***author***>
6:14 2021) <***author***>
into the new data struct (3 weeks ago, Thu Jun 17 15:45:43 2021) <***author***>
uthor***>
4 18:28:50 2021) <***author***>
d May 26 15:31:47 2021) <***author***>
le renaming changes (6 weeks ago, Wed May 26 13:19:12 2021) <***author***>
thor***>
 display changes (6 weeks ago, Tue May 25 10:41:11 2021) <***author***>
ct.*) (6 weeks ago, Mon May 24 19:32:09 2021) <***author***>
) <***author***>
7:18 2021) <***author***>
| | | | | | | | | | | | | | | | | | | | |\ \ \ \ \
ean-up and commenting (7 weeks ago, Mon May 17 17:38:35 2021) <***author***>
:37:24 2021) <***author***>
| | | | | | | | | | | | | | | | | | | | |\ \ \ \ \ \
) <***author***>
02:30 2021) <***author***>
odified the internal data structure, so that syst and tracks conflicting will be:

If, while a middle chunk of a line is displayed, I scroll back up with the up arrow key, and then back down, the top line is duplicated, everything is shifted down one line, but we are still displaying the same middle chunk of the current line of source text.

Working with the down page key seems to work properly

With the page up key it's also proper but limited to the magical width.

With the left and right keys, it's also limited to the magical width (this time with a > at the end of the line) (except when I am all the way to left - there the line extends all the way to my console window's right border, as you would expect, and everything is proper).

I tried setting core.pager to less -R and got the same symptoms.

I then tried less -r and got something almost normal, except that the first few (5 or 10) lines of the tree (usually the most important ones) are missing and only show up if I scroll down and back up with the keys.

This all looks similar to what is described in this post. Fixes involving env -u ROWS and/or env -u COLUMNS have met no success for me.

Thanks in advance.

Charles
  • 988
  • 1
  • 11
  • 28
  • 1
    There has been some discussion of this on the mailing list. It turns out that some combination of Windows software and Git winds up terribly confusing `less` about the actual window size, under some conditions that I do not properly understand. This is being fixed in a future version of Git-for-Windows. – torek Jul 05 '21 at 13:23
  • Thanks, it's a relief to read that. Anywhere I can look at this mailing list, to get such information myself later, esp. regarding the timeframe of fixing this ? I heavily rely on this pager display. – Charles Jul 05 '21 at 14:55
  • 1
    It's archived somewhere (kernel.org? used to be on gmane, but the gmane archives were down the last time I tried them). Google suggests lore.kernel.org (https://www.google.com/search?q=git+mailing+list+archive). – torek Jul 05 '21 at 15:30
  • Oh great, it's archived ! `lore.kernel.org` seems to have recent archives too, but I can't find anything related to this issue. What am I missing ? – Charles Jul 05 '21 at 15:36
  • 1
    Found it [here](https://lore.kernel.org/git/pull.982.git.1623701952823.gitgitgadget@gmail.com/T/) - had to go three pages in. – torek Jul 05 '21 at 16:09
  • Ooooh, there are pages. That's not at all obvious from the visual presentation. Okay, somehow I managed to understand that this issue is considered as closed and a new version will be shipped soon that fixes it. Can't wait ! – Charles Jul 06 '21 at 08:30
  • I used our discussion here to add an answer to the other question; I'll close this one as a duplicate. Or, hm, I would, but my answer isn't upvoted-or-accepted there, so I can't! :-) – torek Jul 06 '21 at 08:47

0 Answers0