The state object is sometimes undefined in the onclick method of the
line by line panel. Because we set it to nil in a bunch of places,
I've decided to just change the main context to 'normal' before setting
it to nil anywhere. That way the keybindings for the line by line panel
won't get executed and we won't get a segfault.
When staging lines (or doing anything that requires the main view to split into two)
we want to split vertically if there's not much width available in the window.
If there is enough width we will split horizontally. The aim here is to allow for
sufficient room in the side panel. We might need to tweak this or make it configurable
but I think it's set to a pretty reasonable default i.e. switching to split vertically
when the window width falls under 220
The issue here was that we were using a string task
but expecting to be able to set the origin straight after
to point at the conflict, but because it's async it was
actually resetting the origin to 0 after a little bit.
The proper solution here is maybe to add a flag to that thing
asking whether you want to reset main's origin. But I'm
too lazy to do that right now so instead I'm just using
setViewContent. That will probably cause issues in the future.
Up till now our approach to rendering things like file diffs, branch logs, and
commit patches, has been to run a command on the command line, wait for it to
complete, take its output as a string, and then write that string to the main
view (or secondary view e.g. when showing both staged and unstaged changes of a
file).
This has caused various issues. For once, if you are flicking through a list of
files and an untracked file is particularly large, not only will this require
lazygit to load that whole file into memory (or more accurately it's equally
large diff), it also will slow down the UI thread while loading that file, and
if the user continued down the list, the original command might eventually
resolve and replace whatever the diff is for the newly selected file.
Following what we've done in lazydocker, I've added a tasks package for when you
need something done but you want it to cancel as soon as something newer comes
up. Given this typically involves running a command to display to a view, I've
added a viewBufferManagerMap struct to the Gui struct which allows you to define
these tasks on a per-view basis.
viewBufferManagers can run files and directly write the output to their view,
meaning we no longer need to use so much memory.
In the tasks package there is a helper method called NewCmdTask which takes a
command, an initial amount of lines to read, and then runs that command, reads
that number of lines, and allows for a readLines channel to tell it to read more
lines. We read more lines when we scroll or resize the window.
There is an adapter for the tasks package in a file called tasks_adapter which
wraps the functions from the tasks package in gui-specific stuff like clearing
the main view before starting the next task that wants to write to the main
view.
I've removed some small features as part of this work, namely the little headers
that were at the top of the main view for some situations. For example, we no
longer show the upstream of a selected branch. I want to re-introduce this in
the future, but I didn't want to make this tasks system too complicated, and in
order to facilitate a header section in the main view we'd need to have a task
that gets the upstream for the current branch, writes it to the header, then
tells another task to write the branch log to the main view, but without
clearing inbetween. So it would get messy. I'm thinking instead of having a
separate 'header' view atop the main view to render that kind of thing (which
can happen in another PR)
I've also simplified the 'git show' to just call 'git show' and not do anything
fancy when it comes to merge commits.
I considered using this tasks approach whenever we write to a view. The only
thing is that the renderString method currently resets the origin of a view and
I don't want to lose that. So I've left some in there that I consider harmless,
but we should probably be just using tasks now for all rendering, even if it's
just strings we can instantly make.
By default, macs have 256 open files allowed by a given process.
This sucks when you end up with over 256 files modified in a repo
because after you've watched all of them, lots of other calls to
the command line will fail due to violating the limit.
Given there's no easy platform agnostic way to see what you've got
configured for how many files a process can have open, I'm going to
arbitrarily set the max to 200 and when we hit the limit we start
unwatching older files to make way for new ones.
WIP
This bug was caused how the timestamp was formatted for the patch file.
On Windows machines, ":" is an invalid character for a filename, but the
`stampNano` format for time contains ":".
This fix adjusts the time format to be the `stampNano` format with "."
subsituted for ":".
when it contains percentages.
This is a really strange one. It's a linting warning in my editor
and it doesn't stop me from compiling, but it breaks `go test`.
A basic file to reproduce what I'm talking about:
package main
import "fmt"
func main() {
notSprintf("test %s") // compiler complains here thinking %s needs a corresponding argument
}
func notSprintf(formatStr string, formatArgs ...interface{}) string {
if formatArgs != nil {
return formatStr
}
return fmt.Sprintf(formatStr, formatArgs...)
}
For this to work you'll need to put this in your ~/.zshrc (or equivalent rc file):
lg()
{
export LAZYGIT_NEW_DIR_FILE=/Users/jesseduffieldduffield/Library/Application\ Support/jesseduffield/lazygit/.lastd
lazygit "$@"
if [ -f $LAZYGIT_NEW_DIR_FILE ]; then
cd "$(cat $LAZYGIT_NEW_DIR_FILE)"
rm -f $LAZYGIT_NEW_DIR_FILE > /dev/null
fi
}
Now we can stage lines by range and we can also stage reversals
meaning we can delete lines or move lines from the working tree
to the index and vice versa.
I looked at how a few different git guis achieved this to iron out
some edge cases, notably ungit and git cola. The end result is
disstinct from both those repos, but I know people care about
licensing and stuff so I'm happy to revisit this if somebody
considers it derivative.
Reusing the `onNewPopupPanel` function to close existing popup panels
(if any) before switching to a new side view. Alse closing any
confirmation prompt.
Previously we were recording output from subprocesses using a multiwriter
and hooking that up to the cmd's stdout to write to both os.Stdout and
a buffer. We would then display the output after the program finished.
This worked well for commands like 'ls' but not for commands like 'vi'
which expect you to be in a tty, and when you've got the cmd's stdout
pointing at a multiwriter, the subprogram thinks we're not in a tty
and then things like terminal corruption can happen. This was the case
with neovim, and even in vim a warning was given with a pause before
starting the program.
Now we're chucking out the multiwriter and instead making it that you
need to press enter after the program has finished to return to lazygit.
This allows you to view the output of the program (e.g. if it's ls) and
then decide that you want to return. It's one level of unnecessary
redirection for editors like vim, but even they could potentially have
output to stderr/stdout that you want to look at before returning.
Please enter the commit message for your changes. Lines starting
Document and use alternative keybinding for generating cheatsheet. Add
alt keybinding fn+up/down for scroll up/down actions.
Also run `go run scripts/generate_cheatsheet.go`.
We were crashing when opening lazygit in a small window because the limit view
was the only view that got created, and there were two functions that referenced
either the 'current' view or the files view, neither of which existed.
Now those functions just return nil if the view does not exist
Return error if the .git exists but is not a directory. This provides a
slightly better failure message for git repo with submodules in case
the '.git' is a file which provides the reference to the parent's .git
folder with the submodule inside.
If your git.skipHookPrefix is set to, say, WIP, in your config, then
hitting 'w' in the files panel will bring up the commit message panel
with 'WIP' pre-filled, so you just need to hit enter to confirm
(or add some more to the message) in order to commit your changes
with the --no-verify flag, meaning the pre-commit hook will be skipped
allows a user to specify a commit message prefix that will tell lazygit to skip
the pre-commit hook. This defaults to WIP. Setting it to the empty string will
disable the feature.
So if my message goes 'WIP: do the thing' then the pre-commit hook will not run
Currently when we want to focus a point on a view (i.e. highlight a
line and ensure it's within the bounds of a view's box, we use the
LinesHeight method on the view to work out how many lines in total
there are.
This is bad because for example if we come back from editing a file,
the view will have no contents so LinesHeight == 0, but we might
be trying to select line 10 because there are actual ten things we
expect to be rendered already. This causes a crash when e.g. 10 is
greater than the height of the view.
So we need to pass in to our FocusPoint method the actual number of
items we want to render, rather than having the method rely on the
LinesHeight, so that the method knows to scroll a bit before setting
the cursor's y position.
Unfortunately this makes for some awkward code with our current setup.
We don't have a good interface type on these state objects so we now
need to explicitly obtain the len() of whatever array we're rendering.
In the case of the menu panel this is even more awkward because the items
list is just an interface{} and it's not easy to get the list of that, so
now when we instantiate a menu we need to pass in the count of items
as well.
The better solution would be to define an interface with a getItems
and getLength method and have all these item arrays become structs
implementing the interface, but I am too lazy to do this right now :)
We already call this function inside the refreshCommitsView function.
We call it there because it's logical that A) one occurs whenever the other does and
B) the commit files only get refreshed after we've updated the commits themselves
