Re: cgit, ages and chronological order

On 2021-Nov-20, at 01:14, Mark Millard <marklmi@yahoo.com> wrote:

> On 2021-Nov-9, at 22:03, Graham Perrin <grahamperrin@gmail.com> wrote:
> 
>> <https://cgit.freebsd.org/src/log/?qt=grep&q=terminfo&h=main>, for example, is it normal for results to _not_ be in chronological order?
>> 
>> (Has it always been so, have I never noticed?)
>> 
> 
> If I gather correctly, a distinction is being made here between searches
> that result in tracking parent-commit order, with parents shown (possibly
> only the "first-parents" path), vs. searches that need not display various
> parent-commits (even omitting some first-parent ones).
> 
> Ones that follow parent-commit order (those involving one of):
> 
> ?qt=grep?q=
> ?qt=author?q=
> ?qt=commiter?q=
> ?qt=range?q=
> ?qt=range?q=SOMETEXT (that is a correct range specification)
> 
> (Those empty ?q= lead to being like https://cgit.freebsd.org/src/log/ lists.)
> 
> Ones that do not follow parent-commit order (because of
> what is omitted): Those with,
> 
> ?qt=grep?q=SOMETEXT
> ?qt=author?q=SOMETEXT
> ?qt=commiter?q=SOMETEXT
> 
> For those last 3, I gather you would like the display order to be by the
> Age order.

Well, those 3 are not complete coverage for cgit: Using
something like,

https://cgit.freebsd.org/src/log/sys/contrib/openzfs

also filters out parent-commit lines and will show things
that are not on a specific branch (at the time). An
example is: 

Revert "Handle partial reads in zfs_read"

where on main the actual patch that covers the issue
shows up in:

zfs: merge openzfs/zfs@6c8f03232 (master) into main

instead and the branch's parent-commit sequencing does
not include the "Revert" one. (The "Revert" one was
a FreeBSD-local commit from what I can tell, pending
the actual openzfs fix.)

(The specific example page happens to show lines sorted
by Age currently. But this may be an accident.)

> I've not figured out what the crtieria is for the order currently displayed
> for these, with your specific example of
> https://cgit.freebsd.org/src/log/?qt=grep&q=terminfo&h=main being
> a good example: The first line after the column title line being:
> 
> * 	Remove parameter names from libzfs.h signatures	Nick Black	2020-01-09	1	-7/+7
> 
> is far from obvious to me.
> 
> (One thing I do not know is how reasonable it would be to sort search
> results that have a huge number of lines to make the end results be in
> Age order.)
> 
> Notes:
> 
> I may not have listed all forms of ?qt=????q=???? that match everything
> or a single subsequence of a parent-commit order vs. those that do not
> match such and so omit some partent-commits. Still, I hope the general
> point is clear.
> 
> For this note, I'm ignoring the "time mismatch" issue. It appears that
> consideration of trying to avoid sizable mismatches is being made.
> (This would not fix already existing sizable mismatches, just prevent
> more.)



===
Mark Millard
marklmi at yahoo.com
( dsl-only.net went
away in early 2018-Mar)