diff mbox series

doc: 'T' status code for git status

Message ID pull.930.git.git.1607501616914.gitgitgadget@gmail.com (mailing list archive)
State Superseded
Headers show
Series doc: 'T' status code for git status | expand

Commit Message

Julien Richard Dec. 9, 2020, 8:13 a.m. UTC
From: Julien Richard <julien.richard@ubisoft.com>

Git status can return 'T' status code which stands for "typechange", fixing the documentation accordingly.

Signed-off-by: Julien Richard <jairbubbles@hotmail.com>
---
    Document 'T' status code for git status
    
    Git status can return 'T' status code which stands for "typechange". I
    can't document more the behavior but it would have helped me a lot to
    see that line in the documentation so I guess it can help others too.

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-git-930%2Fjairbubbles%2FgitStatusTypeChangeCode-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-git-930/jairbubbles/gitStatusTypeChangeCode-v1
Pull-Request: https://github.com/git/git/pull/930

 Documentation/git-status.txt | 1 +
 1 file changed, 1 insertion(+)


base-commit: 3a0b884caba2752da0af626fb2de7d597c844e8b

Comments

Jeff King Dec. 9, 2020, 5:37 p.m. UTC | #1
On Wed, Dec 09, 2020 at 08:13:36AM +0000, Julien Richard via GitGitGadget wrote:

> From: Julien Richard <julien.richard@ubisoft.com>
> 
> Git status can return 'T' status code which stands for "typechange", fixing the documentation accordingly.

Thanks, this is definitely an omission in the documentation.

I wonder if we need a little more, though. The list here:

> --- a/Documentation/git-status.txt
> +++ b/Documentation/git-status.txt
> @@ -197,6 +197,7 @@ codes can be interpreted as follows:
>  * 'R' = renamed
>  * 'C' = copied
>  * 'U' = updated but unmerged
> +* 'T' = type changed
>  
>  Ignored files are not listed, unless `--ignored` option is in effect,
>  in which case `XY` are `!!`.

is followed by a table showing the meaning of those entries in each
slot. Should there be some "T" entries there, too?

I think it could basically show up anywhere that "M" could.

>     Git status can return 'T' status code which stands for "typechange". I
>     can't document more the behavior but it would have helped me a lot to
>     see that line in the documentation so I guess it can help others too.

It is correctly documented in the diff manpage. There we define it as
"change in the type of the file". I'm not sure if that's really making
anything clearer than "type changed". ;)

Perhaps "type changed (e.g., a symbolic link becoming a file)" would be
more descriptive, but I'm not sure it's necessary. (And if so, it
probably would be better placed in the diff documentation).

-Peff
Julien Richard Dec. 9, 2020, 6:01 p.m. UTC | #2
> is followed by a table showing the meaning of those entries in each slot. Should there be some "T" entries there, too?

About the table, I wasn't sure how to fill it so I felt it was safer not to touch it 
Jeff King Dec. 9, 2020, 6:35 p.m. UTC | #3
On Wed, Dec 09, 2020 at 06:01:36PM +0000, Julien Richard wrote:

>  > Perhaps "type changed (e.g., a symbolic link becoming a file)" would be more descriptive
> 
> In fact I'm not sure how get the typechange, I always see the symbolic
> link example but in my case  I have a 100% repro on a repository but
> ZI have no clue why its does so.
> The file is just modified but appears as a type change :-S
> 
> Could it be a bug?

Possibly. If you have a reproduction recipe, share it and we can comment
further.

> About the changes you mentioned should I make the adjustements myself?

Yes, usually you'd squash in any changes you want to make using "commit
--amend" (or "rebase -i", but you only have one patch in the series),
and then you can resubmit via GitGitGadget.

-Peff
Junio C Hamano Dec. 9, 2020, 8:26 p.m. UTC | #4
Jeff King <peff@peff.net> writes:

> I wonder if we need a little more, though. The list here:
>
>> --- a/Documentation/git-status.txt
>> +++ b/Documentation/git-status.txt
>> @@ -197,6 +197,7 @@ codes can be interpreted as follows:
>>  * 'R' = renamed
>>  * 'C' = copied
>>  * 'U' = updated but unmerged
>> +* 'T' = type changed
>>  
>>  Ignored files are not listed, unless `--ignored` option is in effect,
>>  in which case `XY` are `!!`.
>
> is followed by a table showing the meaning of those entries in each
> slot. Should there be some "T" entries there, too?
>
> I think it could basically show up anywhere that "M" could.

Meaning something like the patch attached at the end, perhaps?

> It is correctly documented in the diff manpage. There we define it as
> "change in the type of the file". I'm not sure if that's really making
> anything clearer than "type changed". ;)
>
> Perhaps "type changed (e.g., a symbolic link becoming a file)" would be
> more descriptive, but I'm not sure it's necessary. (And if so, it
> probably would be better placed in the diff documentation).

Sitting next to existing copied, renamed, and updated-but-unmerged,
the extra explanation in parentheses looks a bit out of place.  It
does make sense to have it described in the diff documentation, but
I think it already is covered adequately over there.

Another thing I noticed while looking at the context is that
"updated but" is misleading.  You might be in the middle of a merge,
and the path may have been modified in their history as well as
somewhere in your history, but the change in your history may be far
in the past that you do not even care or remember.  As far as you
are concerned, you didn't update it (e.g. relative to HEAD) at all.
It simply is "unmerged"---if you take a look at the conflicted
contents, saved it and then perhaps "git add" it, you may now be
able to call the path "updated", but at that point it is no longer
"unmerged".

Thanks.


A few things about the attached.

 - it drops "updated but" from the explanation of 'U' in the list.

 - after that, everything in the list becomes a single-word, so
   instead of "type changed", it invents a verb "type-change" and
   uses its pp. form when adding an entry for 'T'.

 - it updates the table to add 'T' next to 'M'.

 - "work tree changed since index" in the table was awkward; it
   rephrases it to "modified in work tree relative to index",
   because (1) these entries are not talking about the working tree
   as a whole; it is one path in the working tree changing its
   type. and (2) using "changed" and "updated" for the same 'M' in
   different context was unnecessarily confusing.  Instead, it uses
   'modified', which appears in the list before the table.
 
 - The "not updated" line was using a leading "\t"ab, and that is
   why the preimage appears unaligned; the patch updates the line to
   use 8 spaces (i.e. ' ' for X and 7 spaces to align).

 Documentation/git-status.txt | 21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)

diff --git c/Documentation/git-status.txt w/Documentation/git-status.txt
index 7731b45f07..2e4b761ecc 100644
--- c/Documentation/git-status.txt
+++ w/Documentation/git-status.txt
@@ -196,7 +196,8 @@ codes can be interpreted as follows:
 * 'D' = deleted
 * 'R' = renamed
 * 'C' = copied
-* 'U' = updated but unmerged
+* 'U' = unmerged
+* 'T' = type-changed
 
 Ignored files are not listed, unless `--ignored` option is in effect,
 in which case `XY` are `!!`.
@@ -204,15 +205,17 @@ in which case `XY` are `!!`.
 ....
 X          Y     Meaning
 -------------------------------------------------
-	 [AMD]   not updated
-M        [ MD]   updated in index
-A        [ MD]   added to index
+        [AMTD]   not changed
+M       [ MTD]   modified in index
+T       [ MTD]   type-changed in index
+A       [ MTD]   added to index
 D                deleted from index
-R        [ MD]   renamed in index
-C        [ MD]   copied in index
-[MARC]           index and work tree matches
-[ MARC]     M    work tree changed since index
-[ MARC]     D    deleted in work tree
+R       [ MTD]   renamed in index
+C       [ MTD]   copied in index
+[MTARC]          index and work tree matches
+[ MTARC]    M    modified in work tree relative to index
+[ MTARC]    T    type-changed in work tree relative to index
+[ MTARC]    D    deleted in work tree
 [ D]        R    renamed in work tree
 [ D]        C    copied in work tree
 -------------------------------------------------
Julien Richard Dec. 9, 2020, 8:56 p.m. UTC | #5
I investigated a bit more my case and it is indeed a file which is modified and is now a symlink. If it's only the main case I think it would be nice to add it to the documentation as the 'typechange' can be quite mysterious otherwise (internet is not much of help on the subject appart from https://code-examples.net/en/q/849dff ).
 
I don't find the table very explicit, for instance I don't understand the []. Is it all the possible statuses in the index / worktree for a given status?
With the T added the table becomes even more complex. As the T status seems like a corner case, mentioning it in the doc briefly would be sufficient, at least for my use case.

Thx,
Julien

-----Message d'origine-----
De : Junio C Hamano <gitster@pobox.com> 
Envoyé : mercredi 9 décembre 2020 21:26
À : Jeff King <peff@peff.net>
Cc : Julien Richard via GitGitGadget <gitgitgadget@gmail.com>; git@vger.kernel.org; Julien Richard <jairbubbles@hotmail.com>; Julien Richard <julien.richard@ubisoft.com>
Objet : Re: [PATCH] doc: 'T' status code for git status

Jeff King <peff@peff.net> writes:

> I wonder if we need a little more, though. The list here:
>
>> --- a/Documentation/git-status.txt
>> +++ b/Documentation/git-status.txt
>> @@ -197,6 +197,7 @@ codes can be interpreted as follows:
>>  * 'R' = renamed
>>  * 'C' = copied
>>  * 'U' = updated but unmerged
>> +* 'T' = type changed
>>  
>>  Ignored files are not listed, unless `--ignored` option is in 
>> effect,  in which case `XY` are `!!`.
>
> is followed by a table showing the meaning of those entries in each 
> slot. Should there be some "T" entries there, too?
>
> I think it could basically show up anywhere that "M" could.

Meaning something like the patch attached at the end, perhaps?

> It is correctly documented in the diff manpage. There we define it as 
> "change in the type of the file". I'm not sure if that's really making 
> anything clearer than "type changed". ;)
>
> Perhaps "type changed (e.g., a symbolic link becoming a file)" would 
> be more descriptive, but I'm not sure it's necessary. (And if so, it 
> probably would be better placed in the diff documentation).

Sitting next to existing copied, renamed, and updated-but-unmerged, the extra explanation in parentheses looks a bit out of place.  It does make sense to have it described in the diff documentation, but I think it already is covered adequately over there.

Another thing I noticed while looking at the context is that "updated but" is misleading.  You might be in the middle of a merge, and the path may have been modified in their history as well as somewhere in your history, but the change in your history may be far in the past that you do not even care or remember.  As far as you are concerned, you didn't update it (e.g. relative to HEAD) at all.
It simply is "unmerged"---if you take a look at the conflicted contents, saved it and then perhaps "git add" it, you may now be able to call the path "updated", but at that point it is no longer "unmerged".

Thanks.


A few things about the attached.

 - it drops "updated but" from the explanation of 'U' in the list.

 - after that, everything in the list becomes a single-word, so
   instead of "type changed", it invents a verb "type-change" and
   uses its pp. form when adding an entry for 'T'.

 - it updates the table to add 'T' next to 'M'.

 - "work tree changed since index" in the table was awkward; it
   rephrases it to "modified in work tree relative to index",
   because (1) these entries are not talking about the working tree
   as a whole; it is one path in the working tree changing its
   type. and (2) using "changed" and "updated" for the same 'M' in
   different context was unnecessarily confusing.  Instead, it uses
   'modified', which appears in the list before the table.
 
 - The "not updated" line was using a leading "\t"ab, and that is
   why the preimage appears unaligned; the patch updates the line to
   use 8 spaces (i.e. ' ' for X and 7 spaces to align).

 Documentation/git-status.txt | 21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)

diff --git c/Documentation/git-status.txt w/Documentation/git-status.txt index 7731b45f07..2e4b761ecc 100644
--- c/Documentation/git-status.txt
+++ w/Documentation/git-status.txt
@@ -196,7 +196,8 @@ codes can be interpreted as follows:
 * 'D' = deleted
 * 'R' = renamed
 * 'C' = copied
-* 'U' = updated but unmerged
+* 'U' = unmerged
+* 'T' = type-changed
 
 Ignored files are not listed, unless `--ignored` option is in effect,  in which case `XY` are `!!`.
@@ -204,15 +205,17 @@ in which case `XY` are `!!`.
 ....
 X          Y     Meaning
 -------------------------------------------------
-	 [AMD]   not updated
-M        [ MD]   updated in index
-A        [ MD]   added to index
+        [AMTD]   not changed
+M       [ MTD]   modified in index
+T       [ MTD]   type-changed in index
+A       [ MTD]   added to index
 D                deleted from index
-R        [ MD]   renamed in index
-C        [ MD]   copied in index
-[MARC]           index and work tree matches
-[ MARC]     M    work tree changed since index
-[ MARC]     D    deleted in work tree
+R       [ MTD]   renamed in index
+C       [ MTD]   copied in index
+[MTARC]          index and work tree matches
+[ MTARC]    M    modified in work tree relative to index
+[ MTARC]    T    type-changed in work tree relative to index
+[ MTARC]    D    deleted in work tree
 [ D]        R    renamed in work tree
 [ D]        C    copied in work tree
 -------------------------------------------------
Jeff King Dec. 9, 2020, 9:32 p.m. UTC | #6
On Wed, Dec 09, 2020 at 12:26:19PM -0800, Junio C Hamano wrote:

> A few things about the attached.
> 
>  - it drops "updated but" from the explanation of 'U' in the list.

Seems reasonable.

>  - after that, everything in the list becomes a single-word, so
>    instead of "type changed", it invents a verb "type-change" and
>    uses its pp. form when adding an entry for 'T'.

Also reasonable.

>  - it updates the table to add 'T' next to 'M'.

Yep, that was what I was thinking would make sense.

>  - "work tree changed since index" in the table was awkward; it
>    rephrases it to "modified in work tree relative to index",
>    because (1) these entries are not talking about the working tree
>    as a whole; it is one path in the working tree changing its
>    type. and (2) using "changed" and "updated" for the same 'M' in
>    different context was unnecessarily confusing.  Instead, it uses
>    'modified', which appears in the list before the table.

The first line seems funny to me now, though (diff abridged):

> -	 [AMD]   not updated
> +        [AMTD]   not changed

If the file is not changed, then why are we even mentioning it? Because
of course it _is_ changed in the filesystem, but the index was not
updated to reflect the change. And that's what I think the original was
getting at with "updated".

TBH, I find the whole table overly confusing. But then, I am completely
comfortable with the notion that it is really showing two diffs, with
their results collated. To me it is simpler to just discuss the two
sides of the diff independently, and then you do not even really need a
table at all ("M" means modified no matter which column it appears in).
But I may not be a representative Git user.

-Peff
Junio C Hamano Dec. 10, 2020, 12:47 a.m. UTC | #7
Jeff King <peff@peff.net> writes:

> TBH, I find the whole table overly confusing. But then, I am completely
> comfortable with the notion that it is really showing two diffs, with
> their results collated. To me it is simpler to just discuss the two
> sides of the diff independently, and then you do not even really need a
> table at all ("M" means modified no matter which column it appears in).
> But I may not be a representative Git user.

Me neither.  I think the folks who originally wanted to see the
table was curious about what combination of X and Y are possible.

For example, if a path is deleted in the index, there won't be any
intelligent thing we can say about the path in the working tree, so
after "git rm COPYING" we'd see "D COPYING", not "DD COPYING", and
after "git rm --cached COPYING && date >COPYING", we'd still see "D
COPYING", not "DM COPYING".

It is unclear how such knowledge would be useful to the readers, so
I won't shed tears if we got rid of the table; the bits saved might
be better used to enhance the description in the list of mnemonics.

Thanks.
diff mbox series

Patch

diff --git a/Documentation/git-status.txt b/Documentation/git-status.txt
index 7731b45f078..850c33b3696 100644
--- a/Documentation/git-status.txt
+++ b/Documentation/git-status.txt
@@ -197,6 +197,7 @@  codes can be interpreted as follows:
 * 'R' = renamed
 * 'C' = copied
 * 'U' = updated but unmerged
+* 'T' = type changed
 
 Ignored files are not listed, unless `--ignored` option is in effect,
 in which case `XY` are `!!`.