threads / patch / 54027

patchgit-gui: accommodate for intent-to-add files

Subject: [PATCH] git-gui: accommodate for intent-to-add files

## tl;dr

7 messages between Aug 12, 2020 and Oct 9, 2020. Diffs are folded; open one to read it.

replies: 6people: 2as markdown or json

Johannes Schindelin via GitGitGadget· Aug 12, 2020, 15:06 UTC · lore
From: Johannes Schindelin <johannes.schindelin@gmx.de>

As of Git v2.28.0, the diff for files staged via `git add -N` marks them as new files. Git GUI was ill-prepared for that, and this patch teaches Git GUI about them.

Please note that this will not even fix things with v2.28.0, as the `rp/apply-cached-with-i-t-a` patches are required on Git's side, too.

This fixes https://github.com/git-for-windows/git/issues/2779
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
    git-gui: accommodate for intent-to-add files
    
    This fixes the intent-to-add bug reported in 
    https://github.com/git-for-windows/git/issues/2779: after a file was
    staged with git add -N, staging hunks/lines would fail silently.
    
    On its own, this patch is not enough, as it requires the patches
    provided in rp/apply-cached-with-i-t-a to be applied on Git's side.
    
    Please note that this patch might need a bit more help, as I do not
    really know whether showing "new file mode 100644" in the diff view is
    desirable, or whether we should somehow try to retain the
    "intent-to-add" state so that unstaging all hunks would return the file
    to "intent-to-add" state.
    
    Thoughts?
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-699%2Fdscho%2Fgit-gui-stage-ita-hunks-and-lines-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-699/dscho/git-gui-stage-ita-hunks-and-lines-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/699
 git-gui.sh   |  2 ++
 lib/diff.tcl | 12 ++++++++----
 2 files changed, 10 insertions(+), 4 deletions(-)
Show changes to 2 files +10 −4

git-gui.sh, lib/diff.tcl

diff --git a/git-gui.sh b/git-gui.sh
index 49bd86e635..e08cb17395 100755
--- a/git-gui.sh
+++ b/git-gui.sh
@@ -2080,6 +2080,7 @@ set all_icons(U$ui_index)   file_merge
 set all_icons(T$ui_index)   file_statechange
 
 set all_icons(_$ui_workdir) file_plain
+set all_icons(A$ui_workdir) file_plain
 set all_icons(M$ui_workdir) file_mod
 set all_icons(D$ui_workdir) file_question
 set all_icons(U$ui_workdir) file_merge
@@ -2106,6 +2107,7 @@ foreach i {
 		{A_ {mc "Staged for commit"}}
 		{AM {mc "Portions staged for commit"}}
 		{AD {mc "Staged for commit, missing"}}
+		{AA {mc "Intended to be added"}}
 
 		{_D {mc "Missing"}}
 		{D_ {mc "Staged for removal"}}
diff --git a/lib/diff.tcl b/lib/diff.tcl
index 871ad488c2..36d3715f7b 100644
--- a/lib/diff.tcl
+++ b/lib/diff.tcl
@@ -582,7 +582,8 @@ proc apply_or_revert_hunk {x y revert} {
 	if {$current_diff_side eq $ui_index} {
 		set failed_msg [mc "Failed to unstage selected hunk."]
 		lappend apply_cmd --reverse --cached
-		if {[string index $mi 0] ne {M}} {
+		set file_state [string index $mi 0]
+		if {$file_state ne {M} && $file_state ne {A}} {
 			unlock_index
 			return
 		}
@@ -595,7 +596,8 @@ proc apply_or_revert_hunk {x y revert} {
 			lappend apply_cmd --cached
 		}
 
-		if {[string index $mi 1] ne {M}} {
+		set file_state [string index $mi 1]
+		if {$file_state ne {M} && $file_state ne {A}} {
 			unlock_index
 			return
 		}
@@ -687,7 +689,8 @@ proc apply_or_revert_range_or_line {x y revert} {
 		set failed_msg [mc "Failed to unstage selected line."]
 		set to_context {+}
 		lappend apply_cmd --reverse --cached
-		if {[string index $mi 0] ne {M}} {
+		set file_state [string index $mi 0]
+		if {$file_state ne {M} && $file_state ne {A}} {
 			unlock_index
 			return
 		}
@@ -702,7 +705,8 @@ proc apply_or_revert_range_or_line {x y revert} {
 			lappend apply_cmd --cached
 		}
 
-		if {[string index $mi 1] ne {M}} {
+		set file_state [string index $mi 1]
+		if {$file_state ne {M} && $file_state ne {A}} {
 			unlock_index
 			return
 		}

base-commit: 469725c1a3d44f7e1475f1d37cd13e0824d4ea41
-- 
gitgitgadget
Pratyush Yadav· Aug 26, 2020, 11:30 UTC · re: Johannes Schindelin via GitGitGadget · lore

Re: [PATCH] git-gui: accommodate for intent-to-add files

Hi Johannes,
Thanks for the patch.
On 12/08/20 03:06PM, Johannes Schindelin via GitGitGadget wrote:
Show 27 quoted lines
> From: Johannes Schindelin <johannes.schindelin@gmx.de>
> 
> As of Git v2.28.0, the diff for files staged via `git add -N` marks them
> as new files. Git GUI was ill-prepared for that, and this patch teaches
> Git GUI about them.
> 
> Please note that this will not even fix things with v2.28.0, as the
> `rp/apply-cached-with-i-t-a` patches are required on Git's side, too.
> 
> This fixes https://github.com/git-for-windows/git/issues/2779
> 
> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> ---
>     git-gui: accommodate for intent-to-add files
>     
>     This fixes the intent-to-add bug reported in 
>     https://github.com/git-for-windows/git/issues/2779: after a file was
>     staged with git add -N, staging hunks/lines would fail silently.
>     
>     On its own, this patch is not enough, as it requires the patches
>     provided in rp/apply-cached-with-i-t-a to be applied on Git's side.
>     
>     Please note that this patch might need a bit more help, as I do not
>     really know whether showing "new file mode 100644" in the diff view is
>     desirable, or whether we should somehow try to retain the
>     "intent-to-add" state so that unstaging all hunks would return the file
>     to "intent-to-add" state.

I built latest Git master (e9b77c84a0) which has `rp/apply-cached-with-i-t-a` and tested this patch. It works... for the most part.

I can select a line set of lines and they get staged/unstaged, which is good. The part that is not good though is that a lot of common operations still don't work as they should:

- I can't click on the icon in the "Unstaged Changes" pane to stage the 
  whole file. Nothing happens when I do that.
- The file that is marked as intent-to-add shows up in both the "Staged 
  Changes" and "Unstaged Changes" panes, with the "Staged Changes" part 
  being empty. Ideally it should only show up in the "Unstaged Changes" 
  pane.
- Selecting the whole file and choosing "Stage Lines for Commit" works 
  well. But choosing "Stage Hunk for Commit" does not. While the changes 
  do get staged, the UI is not properly updated and the file is still 
  listed in the "Unstaged Changes" pane.
  I think the difference here is because for 
  `apply_or_revert_range_or_line`, we call `do_rescan` after it to 
  update the UI, but for `apply_or_revert_hunk` we update the UI 
  "manually" in the function after we are done applying or reverting the 
  changes. So the logic to update the UI needs to be updated to account 
  for this change. Or we can get rid of all that logic and just run a 
  rescan.

And also, like you mentioned, we don't retain the i-t-a state when unstaging. But with some quick testing, I see that Git command line doesn't either (I tried a plain `git restore --staged`). So IMO we should mimic what the command line UI does and not retain the i-t-a state when unstaging.

>     
>     Thoughts?
IMO this is a good start but more work needs to be done before we can 
call this feature finished.
 
Show 73 quoted lines
> Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-699%2Fdscho%2Fgit-gui-stage-ita-hunks-and-lines-v1
> Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-699/dscho/git-gui-stage-ita-hunks-and-lines-v1
> Pull-Request: https://github.com/gitgitgadget/git/pull/699
> 
>  git-gui.sh   |  2 ++
>  lib/diff.tcl | 12 ++++++++----
>  2 files changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/git-gui.sh b/git-gui.sh
> index 49bd86e635..e08cb17395 100755
> --- a/git-gui.sh
> +++ b/git-gui.sh
> @@ -2080,6 +2080,7 @@ set all_icons(U$ui_index)   file_merge
>  set all_icons(T$ui_index)   file_statechange
>  
>  set all_icons(_$ui_workdir) file_plain
> +set all_icons(A$ui_workdir) file_plain
>  set all_icons(M$ui_workdir) file_mod
>  set all_icons(D$ui_workdir) file_question
>  set all_icons(U$ui_workdir) file_merge
> @@ -2106,6 +2107,7 @@ foreach i {
>  		{A_ {mc "Staged for commit"}}
>  		{AM {mc "Portions staged for commit"}}
>  		{AD {mc "Staged for commit, missing"}}
> +		{AA {mc "Intended to be added"}}
>  
>  		{_D {mc "Missing"}}
>  		{D_ {mc "Staged for removal"}}
> diff --git a/lib/diff.tcl b/lib/diff.tcl
> index 871ad488c2..36d3715f7b 100644
> --- a/lib/diff.tcl
> +++ b/lib/diff.tcl
> @@ -582,7 +582,8 @@ proc apply_or_revert_hunk {x y revert} {
>  	if {$current_diff_side eq $ui_index} {
>  		set failed_msg [mc "Failed to unstage selected hunk."]
>  		lappend apply_cmd --reverse --cached
> -		if {[string index $mi 0] ne {M}} {
> +		set file_state [string index $mi 0]
> +		if {$file_state ne {M} && $file_state ne {A}} {
>  			unlock_index
>  			return
>  		}
> @@ -595,7 +596,8 @@ proc apply_or_revert_hunk {x y revert} {
>  			lappend apply_cmd --cached
>  		}
>  
> -		if {[string index $mi 1] ne {M}} {
> +		set file_state [string index $mi 1]
> +		if {$file_state ne {M} && $file_state ne {A}} {
>  			unlock_index
>  			return
>  		}
> @@ -687,7 +689,8 @@ proc apply_or_revert_range_or_line {x y revert} {
>  		set failed_msg [mc "Failed to unstage selected line."]
>  		set to_context {+}
>  		lappend apply_cmd --reverse --cached
> -		if {[string index $mi 0] ne {M}} {
> +		set file_state [string index $mi 0]
> +		if {$file_state ne {M} && $file_state ne {A}} {
>  			unlock_index
>  			return
>  		}
> @@ -702,7 +705,8 @@ proc apply_or_revert_range_or_line {x y revert} {
>  			lappend apply_cmd --cached
>  		}
>  
> -		if {[string index $mi 1] ne {M}} {
> +		set file_state [string index $mi 1]
> +		if {$file_state ne {M} && $file_state ne {A}} {
>  			unlock_index
>  			return
>  		}
> 

These changes look good to me to set up basic functionality. We just need to iron out the rough edges.

> base-commit: 469725c1a3d44f7e1475f1d37cd13e0824d4ea41
> -- 
> gitgitgadget
-- 
Regards,
Pratyush Yadav
Johannes Schindelin· Aug 26, 2020, 07:36 UTC · re: Pratyush Yadav · lore

Re: [PATCH] git-gui: accommodate for intent-to-add files

Hi Pratyush,
On Wed, 26 Aug 2020, Pratyush Yadav wrote:
Show 63 quoted lines
> On 12/08/20 03:06PM, Johannes Schindelin via GitGitGadget wrote:
> > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> >
> > As of Git v2.28.0, the diff for files staged via `git add -N` marks them
> > as new files. Git GUI was ill-prepared for that, and this patch teaches
> > Git GUI about them.
> >
> > Please note that this will not even fix things with v2.28.0, as the
> > `rp/apply-cached-with-i-t-a` patches are required on Git's side, too.
> >
> > This fixes https://github.com/git-for-windows/git/issues/2779
> >
> > Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> > ---
> >     git-gui: accommodate for intent-to-add files
> >
> >     This fixes the intent-to-add bug reported in
> >     https://github.com/git-for-windows/git/issues/2779: after a file was
> >     staged with git add -N, staging hunks/lines would fail silently.
> >
> >     On its own, this patch is not enough, as it requires the patches
> >     provided in rp/apply-cached-with-i-t-a to be applied on Git's side.
> >
> >     Please note that this patch might need a bit more help, as I do not
> >     really know whether showing "new file mode 100644" in the diff view is
> >     desirable, or whether we should somehow try to retain the
> >     "intent-to-add" state so that unstaging all hunks would return the file
> >     to "intent-to-add" state.
>
> I built latest Git master (e9b77c84a0) which has
> `rp/apply-cached-with-i-t-a` and tested this patch. It works... for the
> most part.
>
> I can select a line set of lines and they get staged/unstaged, which is
> good. The part that is not good though is that a lot of common
> operations still don't work as they should:
>
> - I can't click on the icon in the "Unstaged Changes" pane to stage the
>   whole file. Nothing happens when I do that.
>
> - The file that is marked as intent-to-add shows up in both the "Staged
>   Changes" and "Unstaged Changes" panes, with the "Staged Changes" part
>   being empty. Ideally it should only show up in the "Unstaged Changes"
>   pane.
>
> - Selecting the whole file and choosing "Stage Lines for Commit" works
>   well. But choosing "Stage Hunk for Commit" does not. While the changes
>   do get staged, the UI is not properly updated and the file is still
>   listed in the "Unstaged Changes" pane.
>
>   I think the difference here is because for
>   `apply_or_revert_range_or_line`, we call `do_rescan` after it to
>   update the UI, but for `apply_or_revert_hunk` we update the UI
>   "manually" in the function after we are done applying or reverting the
>   changes. So the logic to update the UI needs to be updated to account
>   for this change. Or we can get rid of all that logic and just run a
>   rescan.
>
> And also, like you mentioned, we don't retain the i-t-a state when
> unstaging. But with some quick testing, I see that Git command line
> doesn't either (I tried a plain `git restore --staged`). So IMO we
> should mimic what the command line UI does and not retain the i-t-a
> state when unstaging.

To be quite honest, I had hoped that this might be a good patch to start from... for somebody else (you?)

:-)

Ciao, Dscho

Pratyush Yadav· Aug 26, 2020, 14:52 UTC · re: Johannes Schindelin · lore

Re: [PATCH] git-gui: accommodate for intent-to-add files

On 26/08/20 09:36AM, Johannes Schindelin wrote:
Show 70 quoted lines
> Hi Pratyush,
> 
> On Wed, 26 Aug 2020, Pratyush Yadav wrote:
> 
> > On 12/08/20 03:06PM, Johannes Schindelin via GitGitGadget wrote:
> > > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> > >
> > > As of Git v2.28.0, the diff for files staged via `git add -N` marks them
> > > as new files. Git GUI was ill-prepared for that, and this patch teaches
> > > Git GUI about them.
> > >
> > > Please note that this will not even fix things with v2.28.0, as the
> > > `rp/apply-cached-with-i-t-a` patches are required on Git's side, too.
> > >
> > > This fixes https://github.com/git-for-windows/git/issues/2779
> > >
> > > Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> > > ---
> > >     git-gui: accommodate for intent-to-add files
> > >
> > >     This fixes the intent-to-add bug reported in
> > >     https://github.com/git-for-windows/git/issues/2779: after a file was
> > >     staged with git add -N, staging hunks/lines would fail silently.
> > >
> > >     On its own, this patch is not enough, as it requires the patches
> > >     provided in rp/apply-cached-with-i-t-a to be applied on Git's side.
> > >
> > >     Please note that this patch might need a bit more help, as I do not
> > >     really know whether showing "new file mode 100644" in the diff view is
> > >     desirable, or whether we should somehow try to retain the
> > >     "intent-to-add" state so that unstaging all hunks would return the file
> > >     to "intent-to-add" state.
> >
> > I built latest Git master (e9b77c84a0) which has
> > `rp/apply-cached-with-i-t-a` and tested this patch. It works... for the
> > most part.
> >
> > I can select a line set of lines and they get staged/unstaged, which is
> > good. The part that is not good though is that a lot of common
> > operations still don't work as they should:
> >
> > - I can't click on the icon in the "Unstaged Changes" pane to stage the
> >   whole file. Nothing happens when I do that.
> >
> > - The file that is marked as intent-to-add shows up in both the "Staged
> >   Changes" and "Unstaged Changes" panes, with the "Staged Changes" part
> >   being empty. Ideally it should only show up in the "Unstaged Changes"
> >   pane.
> >
> > - Selecting the whole file and choosing "Stage Lines for Commit" works
> >   well. But choosing "Stage Hunk for Commit" does not. While the changes
> >   do get staged, the UI is not properly updated and the file is still
> >   listed in the "Unstaged Changes" pane.
> >
> >   I think the difference here is because for
> >   `apply_or_revert_range_or_line`, we call `do_rescan` after it to
> >   update the UI, but for `apply_or_revert_hunk` we update the UI
> >   "manually" in the function after we are done applying or reverting the
> >   changes. So the logic to update the UI needs to be updated to account
> >   for this change. Or we can get rid of all that logic and just run a
> >   rescan.
> >
> > And also, like you mentioned, we don't retain the i-t-a state when
> > unstaging. But with some quick testing, I see that Git command line
> > doesn't either (I tried a plain `git restore --staged`). So IMO we
> > should mimic what the command line UI does and not retain the i-t-a
> > state when unstaging.
> 
> To be quite honest, I had hoped that this might be a good patch to start
> from... for somebody else (you?)
I'll take a stab at this during the weekend :-)
-- 
Regards,
Pratyush Yadav
Johannes Schindelin· Oct 9, 2020, 06:56 UTC · re: Pratyush Yadav · lore

Re: [PATCH] git-gui: accommodate for intent-to-add files

Hi Pratyush,
On Wed, 26 Aug 2020, Pratyush Yadav wrote:
Show 72 quoted lines
> On 26/08/20 09:36AM, Johannes Schindelin wrote:
>
> > On Wed, 26 Aug 2020, Pratyush Yadav wrote:
> >
> > > On 12/08/20 03:06PM, Johannes Schindelin via GitGitGadget wrote:
> > > > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> > > >
> > > > As of Git v2.28.0, the diff for files staged via `git add -N` marks them
> > > > as new files. Git GUI was ill-prepared for that, and this patch teaches
> > > > Git GUI about them.
> > > >
> > > > Please note that this will not even fix things with v2.28.0, as the
> > > > `rp/apply-cached-with-i-t-a` patches are required on Git's side, too.
> > > >
> > > > This fixes https://github.com/git-for-windows/git/issues/2779
> > > >
> > > > Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> > > > ---
> > > >     git-gui: accommodate for intent-to-add files
> > > >
> > > >     This fixes the intent-to-add bug reported in
> > > >     https://github.com/git-for-windows/git/issues/2779: after a file was
> > > >     staged with git add -N, staging hunks/lines would fail silently.
> > > >
> > > >     On its own, this patch is not enough, as it requires the patches
> > > >     provided in rp/apply-cached-with-i-t-a to be applied on Git's side.
> > > >
> > > >     Please note that this patch might need a bit more help, as I do not
> > > >     really know whether showing "new file mode 100644" in the diff view is
> > > >     desirable, or whether we should somehow try to retain the
> > > >     "intent-to-add" state so that unstaging all hunks would return the file
> > > >     to "intent-to-add" state.
> > >
> > > I built latest Git master (e9b77c84a0) which has
> > > `rp/apply-cached-with-i-t-a` and tested this patch. It works... for the
> > > most part.
> > >
> > > I can select a line set of lines and they get staged/unstaged, which is
> > > good. The part that is not good though is that a lot of common
> > > operations still don't work as they should:
> > >
> > > - I can't click on the icon in the "Unstaged Changes" pane to stage the
> > >   whole file. Nothing happens when I do that.
> > >
> > > - The file that is marked as intent-to-add shows up in both the "Staged
> > >   Changes" and "Unstaged Changes" panes, with the "Staged Changes" part
> > >   being empty. Ideally it should only show up in the "Unstaged Changes"
> > >   pane.
> > >
> > > - Selecting the whole file and choosing "Stage Lines for Commit" works
> > >   well. But choosing "Stage Hunk for Commit" does not. While the changes
> > >   do get staged, the UI is not properly updated and the file is still
> > >   listed in the "Unstaged Changes" pane.
> > >
> > >   I think the difference here is because for
> > >   `apply_or_revert_range_or_line`, we call `do_rescan` after it to
> > >   update the UI, but for `apply_or_revert_hunk` we update the UI
> > >   "manually" in the function after we are done applying or reverting the
> > >   changes. So the logic to update the UI needs to be updated to account
> > >   for this change. Or we can get rid of all that logic and just run a
> > >   rescan.
> > >
> > > And also, like you mentioned, we don't retain the i-t-a state when
> > > unstaging. But with some quick testing, I see that Git command line
> > > doesn't either (I tried a plain `git restore --staged`). So IMO we
> > > should mimic what the command line UI does and not retain the i-t-a
> > > state when unstaging.
> >
> > To be quite honest, I had hoped that this might be a good patch to start
> > from... for somebody else (you?)
>
> I'll take a stab at this during the weekend :-)

Just a gentle ping: did you get a chance to get this patch into a better shape?

Thanks, Dscho

Pratyush Yadav· Oct 9, 2020, 09:34 UTC · re: Johannes Schindelin · lore

Re: [PATCH] git-gui: accommodate for intent-to-add files

On 09/10/20 08:56AM, Johannes Schindelin wrote:
Show 79 quoted lines
> Hi Pratyush,
> 
> On Wed, 26 Aug 2020, Pratyush Yadav wrote:
> 
> > On 26/08/20 09:36AM, Johannes Schindelin wrote:
> >
> > > On Wed, 26 Aug 2020, Pratyush Yadav wrote:
> > >
> > > > On 12/08/20 03:06PM, Johannes Schindelin via GitGitGadget wrote:
> > > > > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> > > > >
> > > > > As of Git v2.28.0, the diff for files staged via `git add -N` marks them
> > > > > as new files. Git GUI was ill-prepared for that, and this patch teaches
> > > > > Git GUI about them.
> > > > >
> > > > > Please note that this will not even fix things with v2.28.0, as the
> > > > > `rp/apply-cached-with-i-t-a` patches are required on Git's side, too.
> > > > >
> > > > > This fixes https://github.com/git-for-windows/git/issues/2779
> > > > >
> > > > > Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> > > > > ---
> > > > >     git-gui: accommodate for intent-to-add files
> > > > >
> > > > >     This fixes the intent-to-add bug reported in
> > > > >     https://github.com/git-for-windows/git/issues/2779: after a file was
> > > > >     staged with git add -N, staging hunks/lines would fail silently.
> > > > >
> > > > >     On its own, this patch is not enough, as it requires the patches
> > > > >     provided in rp/apply-cached-with-i-t-a to be applied on Git's side.
> > > > >
> > > > >     Please note that this patch might need a bit more help, as I do not
> > > > >     really know whether showing "new file mode 100644" in the diff view is
> > > > >     desirable, or whether we should somehow try to retain the
> > > > >     "intent-to-add" state so that unstaging all hunks would return the file
> > > > >     to "intent-to-add" state.
> > > >
> > > > I built latest Git master (e9b77c84a0) which has
> > > > `rp/apply-cached-with-i-t-a` and tested this patch. It works... for the
> > > > most part.
> > > >
> > > > I can select a line set of lines and they get staged/unstaged, which is
> > > > good. The part that is not good though is that a lot of common
> > > > operations still don't work as they should:
> > > >
> > > > - I can't click on the icon in the "Unstaged Changes" pane to stage the
> > > >   whole file. Nothing happens when I do that.
> > > >
> > > > - The file that is marked as intent-to-add shows up in both the "Staged
> > > >   Changes" and "Unstaged Changes" panes, with the "Staged Changes" part
> > > >   being empty. Ideally it should only show up in the "Unstaged Changes"
> > > >   pane.
> > > >
> > > > - Selecting the whole file and choosing "Stage Lines for Commit" works
> > > >   well. But choosing "Stage Hunk for Commit" does not. While the changes
> > > >   do get staged, the UI is not properly updated and the file is still
> > > >   listed in the "Unstaged Changes" pane.
> > > >
> > > >   I think the difference here is because for
> > > >   `apply_or_revert_range_or_line`, we call `do_rescan` after it to
> > > >   update the UI, but for `apply_or_revert_hunk` we update the UI
> > > >   "manually" in the function after we are done applying or reverting the
> > > >   changes. So the logic to update the UI needs to be updated to account
> > > >   for this change. Or we can get rid of all that logic and just run a
> > > >   rescan.
> > > >
> > > > And also, like you mentioned, we don't retain the i-t-a state when
> > > > unstaging. But with some quick testing, I see that Git command line
> > > > doesn't either (I tried a plain `git restore --staged`). So IMO we
> > > > should mimic what the command line UI does and not retain the i-t-a
> > > > state when unstaging.
> > >
> > > To be quite honest, I had hoped that this might be a good patch to start
> > > from... for somebody else (you?)
> >
> > I'll take a stab at this during the weekend :-)
> 
> Just a gentle ping: did you get a chance to get this patch into a better
> shape?

No, I have not. I'll try my hand on it tomorrow and try to get it done in time for 2.29.

-- 
Regards,
Pratyush Yadav
Johannes Schindelin· Oct 9, 2020, 13:22 UTC · re: Pratyush Yadav · lore

Re: [PATCH] git-gui: accommodate for intent-to-add files

Hi Pratyush,
On Fri, 9 Oct 2020, Pratyush Yadav wrote:
Show 82 quoted lines
> On 09/10/20 08:56AM, Johannes Schindelin wrote:
> >
> > On Wed, 26 Aug 2020, Pratyush Yadav wrote:
> >
> > > On 26/08/20 09:36AM, Johannes Schindelin wrote:
> > >
> > > > On Wed, 26 Aug 2020, Pratyush Yadav wrote:
> > > >
> > > > > On 12/08/20 03:06PM, Johannes Schindelin via GitGitGadget wrote:
> > > > > > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> > > > > >
> > > > > > As of Git v2.28.0, the diff for files staged via `git add -N` marks them
> > > > > > as new files. Git GUI was ill-prepared for that, and this patch teaches
> > > > > > Git GUI about them.
> > > > > >
> > > > > > Please note that this will not even fix things with v2.28.0, as the
> > > > > > `rp/apply-cached-with-i-t-a` patches are required on Git's side, too.
> > > > > >
> > > > > > This fixes https://github.com/git-for-windows/git/issues/2779
> > > > > >
> > > > > > Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> > > > > > ---
> > > > > >     git-gui: accommodate for intent-to-add files
> > > > > >
> > > > > >     This fixes the intent-to-add bug reported in
> > > > > >     https://github.com/git-for-windows/git/issues/2779: after a file was
> > > > > >     staged with git add -N, staging hunks/lines would fail silently.
> > > > > >
> > > > > >     On its own, this patch is not enough, as it requires the patches
> > > > > >     provided in rp/apply-cached-with-i-t-a to be applied on Git's side.
> > > > > >
> > > > > >     Please note that this patch might need a bit more help, as I do not
> > > > > >     really know whether showing "new file mode 100644" in the diff view is
> > > > > >     desirable, or whether we should somehow try to retain the
> > > > > >     "intent-to-add" state so that unstaging all hunks would return the file
> > > > > >     to "intent-to-add" state.
> > > > >
> > > > > I built latest Git master (e9b77c84a0) which has
> > > > > `rp/apply-cached-with-i-t-a` and tested this patch. It works... for the
> > > > > most part.
> > > > >
> > > > > I can select a line set of lines and they get staged/unstaged, which is
> > > > > good. The part that is not good though is that a lot of common
> > > > > operations still don't work as they should:
> > > > >
> > > > > - I can't click on the icon in the "Unstaged Changes" pane to stage the
> > > > >   whole file. Nothing happens when I do that.
> > > > >
> > > > > - The file that is marked as intent-to-add shows up in both the "Staged
> > > > >   Changes" and "Unstaged Changes" panes, with the "Staged Changes" part
> > > > >   being empty. Ideally it should only show up in the "Unstaged Changes"
> > > > >   pane.
> > > > >
> > > > > - Selecting the whole file and choosing "Stage Lines for Commit" works
> > > > >   well. But choosing "Stage Hunk for Commit" does not. While the changes
> > > > >   do get staged, the UI is not properly updated and the file is still
> > > > >   listed in the "Unstaged Changes" pane.
> > > > >
> > > > >   I think the difference here is because for
> > > > >   `apply_or_revert_range_or_line`, we call `do_rescan` after it to
> > > > >   update the UI, but for `apply_or_revert_hunk` we update the UI
> > > > >   "manually" in the function after we are done applying or reverting the
> > > > >   changes. So the logic to update the UI needs to be updated to account
> > > > >   for this change. Or we can get rid of all that logic and just run a
> > > > >   rescan.
> > > > >
> > > > > And also, like you mentioned, we don't retain the i-t-a state when
> > > > > unstaging. But with some quick testing, I see that Git command line
> > > > > doesn't either (I tried a plain `git restore --staged`). So IMO we
> > > > > should mimic what the command line UI does and not retain the i-t-a
> > > > > state when unstaging.
> > > >
> > > > To be quite honest, I had hoped that this might be a good patch to start
> > > > from... for somebody else (you?)
> > >
> > > I'll take a stab at this during the weekend :-)
> >
> > Just a gentle ping: did you get a chance to get this patch into a better
> > shape?
>
> No, I have not. I'll try my hand on it tomorrow and try to get it done
> in time for 2.29.
Thank you so much!

Ciao, Dscho

← back to recent threads