On 2026-01-26 18:44 (-4), Bill Pragnell wrote:
> Jim Henderson <nos### [at] nospamcom> wrote:
>> On Sat, 24 Jan 2026 23:56:33 -0400, Cousin Ricky wrote:
>>
>>> The files README.md, gemcuts.pov, and gemcuts_description.txt were
>>> modified while one branch was checked out, and the changes show in the
>>> other branch as well.
>>
>> Hmm.  I guess then we'd need to see what the status looks like from each
>> branch.  What you're seeing isn't consistent with my experience, at least.
> 
> I think that's what I would expect. If the changes don't have conflicts with any
> of the other branches then switching branches makes no difference, the changes
> will be left alone (unless there is a conflict, then git will warn you to commit
> your changes or lose them when switching branch). If you commit the changes to a
> branch, then you'll only see those changes on that branch - and now they won't
> be listed as changes by 'git status' any more.

This is exactly what I've seen since Sunday.

> An untracked file is just a big change after all.

Post a reply to this message

On 2026-01-26 02:00 (-4), Jim Henderson wrote:
> On Mon, 26 Jan 2026 17:44:27 EST, Bill Pragnell wrote:
>>
>> An untracked file is just a big change after all.
> 
> But it looks like the file is tracked, so any changes to it made in one 
> branch shouldn't affect other branches, unless they're either not tracked 
> (which isn't the case) or in .gitignore (which would *probably* require it 
> to have been explicitly added in some way, and I expect CR would know if 
> he'd done that).

I'm sensing disagreement in what "tracked" means.  Jim is using
"tracked" the way I've understood it, but Bill is describing the
behavior I'm seeing from git.

But .gitignore is one aspect of git that has never given me any surprises.

Post a reply to this message

On Tue, 27 Jan 2026 10:02:14 -0400, Cousin Ricky wrote:

> On 2026-01-26 02:00 (-4), Jim Henderson wrote:
>> On Mon, 26 Jan 2026 17:44:27 EST, Bill Pragnell wrote:
>>>
>>> An untracked file is just a big change after all.
>> 
>> But it looks like the file is tracked, so any changes to it made in one
>> branch shouldn't affect other branches, unless they're either not
>> tracked (which isn't the case) or in .gitignore (which would *probably*
>> require it to have been explicitly added in some way, and I expect CR
>> would know if he'd done that).
> 
> I'm sensing disagreement in what "tracked" means.  Jim is using
> "tracked" the way I've understood it, but Bill is describing the
> behavior I'm seeing from git.
> 
> But .gitignore is one aspect of git that has never given me any
> surprises.

Can you describe the workflow you're using, step by step - commands 
executed, and so on - in order to try to reproduce what you're seeing?


-- 
"I learned long ago, never to wrestle with a pig. You get dirty, and 
besides, the pig likes it." - George Bernard Shaw

Post a reply to this message

Jim Henderson <nos### [at] nospamcom> wrote:
> On Tue, 27 Jan 2026 10:02:14 -0400, Cousin Ricky wrote:
> > On 2026-01-26 02:00 (-4), Jim Henderson wrote:
> >> But it looks like the file is tracked, so any changes to it made in one
> >> branch shouldn't affect other branches,

Are we confusing our usage of 'change' perhaps? A 'change' to a file (in
git-speak) refers to an edit, without any git interaction, which is what I've
been assuming. A change to a branch is usually called a 'commit', and that will
indeed be unique to the branch.

> > I'm sensing disagreement in what "tracked" means.  Jim is using
> > "tracked" the way I've understood it, but Bill is describing the
> > behavior I'm seeing from git.

I think git itself may also be confusing the issue here. 'Tracked' just means
the file has been added to the repo. If I add a new file to a project (without
git interaction - just make a new file within a repo), it is listed as
'untracked' because it does not yet have an entry in the repo - its entire
contents are a 'change'. If I 'git add' and then 'git commit' that file, it is
then 'tracked' as part of that branch from that point on. But you could say that
even an 'untracked' file is tracked in the sense of git being aware of it,
because it appears in the status. Only files/directories in .gitignore are truly
untracked because edits to them will be ignored.

Uncommitted changes (i.e. edits to a file) will not be affected by switching
branches, unless there is a conflict - i.e. if the edits apply to a section of a
file that is different in the two branches. In that case, git will not switch
branches, but instead tell you to commit or 'stash' your changes before
switching (stashing is saving your changes on a temp stack rather than
committing them).

Sorry if I'm muddying the water here! I remember being quite confused about how
git worked when I first started using it.

Bill

Post a reply to this message

On Fri, 30 Jan 2026 21:42:09 EST, Bill Pragnell wrote:

> Jim Henderson <nos### [at] nospamcom> wrote:
>> On Tue, 27 Jan 2026 10:02:14 -0400, Cousin Ricky wrote:
>> > On 2026-01-26 02:00 (-4), Jim Henderson wrote:
>> >> But it looks like the file is tracked, so any changes to it made in
>> >> one branch shouldn't affect other branches,
> 
> Are we confusing our usage of 'change' perhaps? A 'change' to a file (in
> git-speak) refers to an edit, without any git interaction, which is what
> I've been assuming. A change to a branch is usually called a 'commit',
> and that will indeed be unique to the branch.

No, that's what I'm calling a change - an edit to the file.

>> > I'm sensing disagreement in what "tracked" means.  Jim is using
>> > "tracked" the way I've understood it, but Bill is describing the
>> > behavior I'm seeing from git.
> 
> I think git itself may also be confusing the issue here. 'Tracked' just
> means the file has been added to the repo. If I add a new file to a
> project (without git interaction - just make a new file within a repo),
> it is listed as 'untracked' because it does not yet have an entry in the
> repo - its entire contents are a 'change'. If I 'git add' and then 'git
> commit' that file, it is then 'tracked' as part of that branch from that
> point on. But you could say that even an 'untracked' file is tracked in
> the sense of git being aware of it, because it appears in the status.
> Only files/directories in .gitignore are truly untracked because edits
> to them will be ignored.

Hmmm.  I just created a quick test repo, and it's not working the way I 
remembered it, but it is working the way you're describing and CR is 
saying is happening for his setup, too.

> Uncommitted changes (i.e. edits to a file) will not be affected by
> switching branches, unless there is a conflict - i.e. if the edits apply
> to a section of a file that is different in the two branches. In that
> case, git will not switch branches, but instead tell you to commit or
> 'stash' your changes before switching (stashing is saving your changes
> on a temp stack rather than committing them).

This is what I'm used to seeing.  But my usage generally had me making 
changes within a branch, not creating the branch myself, and then the 
developer who created the branch would merge things back (my main use case 
for this was years ago creating in-product text strings for documentation 
purposes - tooltips and the like).

I've used it a fair bit on my own since then, but generally haven't needed 
to branch since the stuff I'm using it for is all for my own amusement.

> Sorry if I'm muddying the water here! I remember being quite confused
> about how git worked when I first started using it.
> 
> Bill

No worries, you got me to actually try it and see, and it wasn't matching 
my memory.  Show now I need to figure out why my memory is different than 
what I actually saw in my simple test.



-- 
"I learned long ago, never to wrestle with a pig. You get dirty, and 
besides, the pig likes it." - George Bernard Shaw

Post a reply to this message

On 31 Jan 2026 00:52:22 -0500, Jim Henderson wrote:

> This is what I'm used to seeing.  But my usage generally had me making
> changes within a branch, not creating the branch myself, and then the
> developer who created the branch would merge things back (my main use
> case for this was years ago creating in-product text strings for
> documentation purposes - tooltips and the like).

I think I understand better now - once I committed the changes to master, 
then the branch I created (test1) "reverted" to what was originally in the 
file when I branched it (ie, it was empty).



-- 
"I learned long ago, never to wrestle with a pig. You get dirty, and 
besides, the pig likes it." - George Bernard Shaw

Post a reply to this message

On 31 Jan 2026 00:55:13 -0500, Jim Henderson wrote:

> On 31 Jan 2026 00:52:22 -0500, Jim Henderson wrote:
> 
>> This is what I'm used to seeing.  But my usage generally had me making
>> changes within a branch, not creating the branch myself, and then the
>> developer who created the branch would merge things back (my main use
>> case for this was years ago creating in-product text strings for
>> documentation purposes - tooltips and the like).
> 
> I think I understand better now - once I committed the changes to
> master, then the branch I created (test1) "reverted" to what was
> originally in the file when I branched it (ie, it was empty).

So it seems to be that once a file is modified and then committed to a 
branch, then changes to it will require a stash or commit before changing 
branches again.

I probably never ran into this since the files I was modifying had already 
been through many iterations by the time I saw them.  But it makes logical 
sense to me that each branch isn't a full copy of everything in the state 
at which it was first added, and that it would be consistent across 
branches until it was changed in one of them.

I'd guess that if a change were committed in 'branch1' and you switched 
back to 'master' where it wasn't committed, and created a new branch 
'branch2', that the behavior would be consistent as well, at least until 
'branch1' was merged back into 'master'.



-- 
"I learned long ago, never to wrestle with a pig. You get dirty, and 
besides, the pig likes it." - George Bernard Shaw

Post a reply to this message