test: Expected a combining operator like '-a' at index 4

[joehillen]

Describe the bug Getting this error:

test: Expected a combining operator like '-a' at index 4                                  = files _fifc_source_files _fifc_comp_2 _fifc_comp_3 _fifc_comp_4 _fifc_comp_5 _fifc_comp_6 _fifc_comp_7 _fifc_comp_8[1]
         ^
~/.config/fish/functions/_fifc_action.fish (line 1):
in function '_fifc_action' with arguments 'source'
    called on line 1 of file ~/.config/fish/functions/_fifc.fish
in command substitution
    called on line 25 of file ~/.config/fish/functions/_fifc.fish

This happens right before fzf launches, by pressing ctrl-x.

The stack trace isn't exactly helpful. Any ideas? What is it trying to do? I can probably figure it out myself if you can give me a tip.

[gazorby]

Hi @joehillen!

What was the command you were typing? Did you add any custom completion rules?

Completion rules are stored in lists and are tested against the command you are typing. As _fifc_action is called with source arg, the test call may come from the command evaluated here: https://github.com/gazorby/fifc/blob/adff5966739667d4c13d6388372e40f821571208/functions/_fifc_action.fish#L58 This evaluated command (defined in completion rule) should retrieve a list of completion suitable for the current context, and the error you see happens when evaluating that particular command.

[Subtixx]

This error happens for me when there's no manpage for the particular command.

[gazorby]

Strangely, I can reproduce but only when using WSL, it works fine on Linux distros

[Subtixx]

I'm using arch 6.0.7-arch1-1 and it shows me the error

[timcleaver]

I am not sure why it is happening, but for me this came about when using fish within tmux and not within an xterm. So, it seemed like it was a double initialisation issue. The other reason to make me think this is that when I experience the error various _fifc_comp_ variables had duplicated content. So the way I fixed it for me was to delete the _fifc_comp_ variable before setting it. I did this with the following diff:

diff --git a/functions/fifc.fish b/functions/fifc.fish
index f7c5910..3f81edb 100644
--- a/functions/fifc.fish
+++ b/functions/fifc.fish
@@ -38,6 +38,8 @@ function fifc -d "Add your own fish fzf completions"

     set _fifc_comp_count (math $_fifc_comp_count + 1)
     set -l count $_fifc_comp_count
+    set -e -U "_fifc_comp_$count"
+    set -e -g "_fifc_comp_$count"
     set -Ux "_fifc_comp_$count"
     set -a "_fifc_comp_$count" "$_flag_n"
     set -a "_fifc_comp_$count" "$_flag_r"

Without the fix I would get:

_fifc_comp_10=test "$fifc_group" = processes    ps -ax -o pid=,command=   test "$fifc_group" = processes    ps -ax -o pid=,command=

With the fix:

_fifc_comp_10=test "$fifc_group" = processes    ps -ax -o pid=,command=

Which is at least valid syntax. I am not sure this completely resolves all the issues with the _fifc_comp_* variables. Because when in tmux I still get a bunch of duplicates:

> env | sort | grep _fifc | grep "processes\s*ps -ax"
_fifc_comp_10=test "$fifc_group" = processes    ps -ax -o pid=,command=
_fifc_comp_18=test "$fifc_group" = processes    ps -ax -o pid=,command=
_fifc_comp_26=test "$fifc_group" = processes    ps -ax -o pid=,command=
_fifc_comp_2=test "$fifc_group" = processes    ps -ax -o pid=,command=
_fifc_comp_34=test "$fifc_group" = processes    ps -ax -o pid=,command=

I suspect the fifc function needs to also check whether there is already a variable with the matching "body" and only add new ones. But my above hack did enough to solve the issue for me.

[gazorby]

Good catch @timcleaver! Should be gone on main now, that fixed for me too ;) Fixed on main