When I run gvim from MSYS, things go wrong during initialization. Namely, gvim can't find the initialization files that are in 'C:\Documents and Settings\username\vimfiles.
[Specifically, gvim reports the error E117: Unknown function: pathogen#infect during initialization, so it never found autoload\pathogen.vim. Doing :scriptnames also confirms that none of the setup files from vimfiles\ are run.]
I think I've debugged why it goes wrong. When you start MSYS, the MSYS shell inherits the windows enviroment variables, but changes some of them to it's custom values. C:\Documents and Settings\username is the value of $HOME in Windows, but MSYS sets it to something like C:/msys/user name. And of course, Vim uses $HOME to find the right initialization files.
I also notice set shell? has changed to something like shell=C:/msys/bin/sh instead of shell=C:\WINDOWS\system32\cmd.exe, but I hope this isn't important for fixing the initialization problem.
I need to run gvim with the normal windows environment variables. At least I need to be able to manually override a few important ones like $HOME to something I specify (i.e., I'm not concerned about my windows $HOME changing, so it's fine to use a static value).
I tried to reset $HOME manually in my vimrc, but by then it is too late.
Is there some trick to specifying $HOME early on during initialization, or as an extra command line parameter?
Alternatively, is there some trick with running commands from msys differently? I know almost nothing about how the shell C:/msys/bin/sh works, but I could conceive of some extra arguments that changes the visibile environment for the command (e.g. gvim.exe) you are typing.
---Edit---
Reposting the solution that worked (it achieves the later idea):
Instead of running gvim.exe, run the command HOME="C:\Documents and Settings\username" gvim.exe
In bash and other UNIX shells, you can do:
$ HOME='/path/to/dir' gvim
to temporarily set $HOME to a different value.
I admit I'm not familiar at all with the Windows command line, but it might be worth a try.
I believe you can define $HOME just like any other environment variable.
Try adding this to the start of your .vimrc.
let $HOME="C:\Documents and Settings\username"
I had a similar issue when running Cygwin (which is similar to MSYS).
The easiest solution for me was to simply set the HOME environment variable to an empty string. Otherwise, the Cygwin HOME value would be appended to the Windows USERPROFILE in GVim and it would fail to start correctly. Not explicitly setting the full path of the HOME means that I can use the same start-up files on different systems where the USERNAME may not be the same.
Likewise, the SHELL environment variable should not be inherited from Cygwin Bash. This results in errors when running shell or external programs via !. Again, I set this to be an empty variable since Vim is smart enough to figure out what it should be.
In my .bashrc, I set the following alias for running Windows gvim which starts GVim with the HOME and SHELL variables set to an empty string – just for that command.
gvim="/cygdrive/c/Program Files (x86)/Vim/vim74/gvim"
if [ -x "$gvim" ]; then
alias gvim="HOME= SHELL= \"$gvim\""
fi
unset gvim
I achieved this by making a windows symbol link (Win 7 or higher).
Via a symbol link, you can even make vim and gvim to use the same configuration and plugin.
Related
This is bothering me more than it should and has me completely stumped. I feel like like finding the answer will have some good learning opportunities so hopefully it's relevant.
I do embedded C development with Vim and have a setup for hobbyist stuff with Arduino (using Arduino Makefile). I use :make with some shortcuts with build projects.
An external define resolves the Arduino Makefile root directory in the project level Makefile: 'ARDMK_DIR=/usr/local/opt/arduino-mk'. This is define as an export in my shell (zsh). This is where it gets weird:
Using make at the shell prompt the project builds fine:
make -d
This program built for i386-apple-darwin11.3.0
Reading makefiles...
Reading makefile `Makefile'...
Reading makefile `/usr/local/opt/arduino-mk/Arduino.mk' (search path) (no ~ expansion)...
However using :make in Vim the define becomes something from an old install:
:make
This program built for i386-apple-darwin11.3.0
Reading makefiles...
Reading makefile `Makefile'...
Reading makefile `/usr/local/Cellar/arduino-mk/1.5.2/Arduino.mk' (search path) (no ~ expansion)...
Makefile:24: /usr/local/Cellar/arduino-mk/1.5.2/Arduino.mk: No such file or directory
I cannot for the life of me find where ARDMK_DIR is being re-defined to '/usr/local/Cellar/arduino-mk/1.5.2'. Things I have tried:
setlocal makeprg=echo\ $ARDMK_DIR\ &&\ make\ -d\: echo comes back with my shell define (/usr/local/opt/arduino-mk), but make fails with the error above!!
:echo $ARDMK_DIR: again returns my shell define.
ag my home directory for ARDMK_DIR, the only place it is defined is in my shell exports. Did since for my root directory to and same thing. Same thing for $VIMRUNTIME
Even vim-disptach works fine calling the same makeprg?!
Re-define ARDMK_DIR in the project Makefile. Everything builds find as expected. I don't want to do this however as I compile with different systems.
The same vim config works on other macOS and Linux systems with expected behaviour.
Some where between echo and the actual execution of make, ARDMK_DIR is being re-defined. Why and can anyone think of a way of finding out where and solving this?
Zsh has multiple init files that are sourced. The file .zshenv is always sourced, when the shell starts and the file .zshrc is only sourced when the shell is started in interactive mode.
If you define the variable ARDMK_DIR with different values in .zshenv and in .zshrc, the value from .zshrc will be used when you work interactive with the shell (entering commands, starting Vim, ...).
But when Vim starts a command it will start a non-interactive shell. In that case only the file .zshenv will be sourced, so you get the value from that file.
One question left:
Why did the following command first echo the correct value, but make uses the wrong?
:setlocal makeprg=echo\ $ARDMK_DIR\ &&\ make\ -d\
For testing, I started Vim under strace. Then :
:set makeprg=echo\ $EDITOR
:make
In the strace file I found the following line:
execve("/usr/bin/zsh", ["/usr/bin/zsh", "-c", "echo vi 2>&1| tee /tmp/vdxR5DH/"...], [/* 86 vars */]) = 0
As you can see, Vim executes echo vi, so it already expanded the environment variable $EDITOR to its value before calling the shell.
So the answer to the above question is, that the echo command echos the text, that Vim inserted into the command line while the make command gets the the variable value from the environment. As it is a non-interactive shell, it is the value from .zshenv.
This question is specific to using cmake as part of Android Studio build process.
I'm using Android Studio 2.2 preview 7 on linux (ubuntu)
Inside the CMakeLists.txt I am able to access the Android NDK path using: ${ANDROID_NDK}
But how can I access:
Any environment variable ?
If not possible, at least the Android SDK path ?
I already tried to used $ENV{name_of_the_environment_variable_here} but it's always empty, even if the environment variable exist. I guess that when gradle invoke cmake it "hide" the env var somehow.
I don't think you can use $ENV, it's just an example of a variable because they're environment variables. However, you should be able to type env and hit enter for a list of the variables you currently have set. Then, the ones you see in the list, you can invoke by typing $VARIABLE_NAME, using a command before them to get them to do something. E.g. echo $VARIABLE will echo your variable to stdout.
I'm not sure how $ANDROID_SDK was set, if it was part of an install process, etc. but generally you would set user environment variables in .profile, .bash_profile or .bashrc configuration files. These files are read by the shell in that order. System-wide variables are set in /etc/environment, /etc/profile, and /etc/bash.bashrc, but you probably don't want to mess with those (most distros encourage making ancillary additions in /etc/profile.local, but that's a story for another answer).
It doesn't particularly matter which one of these files you use, unless what you're trying to do interacts with the order in which they are loaded. Generally, I look for where the variables have been set by either the OS or other stuff I've added and put them near those. You can find where environment variables are set by typing:
% for i in .profile .bash_profile .bashrc; do grep -H PATH $HOME/$i; done
(% is the prompt, don't type %)
.. and this will loop through the 3 files and show you if a user $PATH is set in any of them.
Bash uses the export ENV command as opposed to set ENV, which is from the original sh, which AFAIK is only default on FreeBSD and derivatives like pfSense anymore. Almost all other OS use Bash by default, except MacOS which recently moved to zsh and also uses export, and OpenBSD which uses ksh (nobody uses OpenBSD).
If you want to verify which shell you are using, type echo $SHELL, or echo $0 and hit enter, and it should let you know.
You can add the environment variable ephemerally by typing this command in your bash terminal and pressing enter:
% export ANDROID_SDK_ROOT=/home/username/AndroidSDK
To be clear, this is an example path, so it'd be best to use the actual path in which your android SDK files reside. However, this example was a default install location Android Studio tried to use when I installed it recently, so if you're not sure where they are, it's probably a good place to check.
To have a more permanent setting of your environment variable, open a text editor and add the line above to one of the configuration files I mentioned in the first paragraph (they'll be in your $HOME folder). Or, you can run this from the prompt and it'll add it to your file automatically:
% echo 'export ANDROID_SDK_ROOT=/home/username/AndroidSDK' >> $HOME/.bashrc
Take care to use two angle brackets and not one, as one angle bracket will overwrite the entire file with the single line.
How can I access Any environment variable ?
If you're not sure which folder is $HOME, try typing cd $HOME and hitting enter - that'll take you there. That's how you access environment variables - use a command with the invocation of the variable and it should act as if you had typed out the entire thing.
To access environment variables, type echo $NAME_OF_VARIABLE and it should echo it to the screen. If you want to search your three config files I mentioned in the beginning for where an environment variable is set, you can use grep as I did earlier, just changing the search string for whatever you're looking for. E.g. (while in $HOME):
% grep SDK_ROOT .bashrc .profile .bash_profile
Or you can type env to list all the currently set variables and filter them by piping the output to the grep command:
% env | grep SDK
If you want to just list all of the set variables and root around the entire thing, just type env instead of piping it to grep (grep's a filter).
Lastly, I'll give you an example of my $ANDROID_SDK_ROOT $ANDROID_SDK and $SDK_ROOT variables in my .bashrc - I noticed while installing these tools, they use all three (isn't that fun?):
% grep ANDROID .bashrc
export ANDROID_SDK=$HOME/development/Android/SDK
export ANDROID_SDK_ROOT=$HOME/development/Android/SDK
export PATH=$PATH:$ANDROID_SDK:$JAVA_HOME:$ANDROID_SDK/cmdline-tools/latest/bin:$ANDROID_SDK/build-tools/32.0.0:$ANDROID_SDK/emulator:$ANDROID_SDK/emulator/bin64:$ANDROID_SDK/tools:$ANDROID_SDK/tools/bin:$ANDROID_SDK/extras:$ANDROID_SDK/platform-tools:$HOME/development/AndroidStudio/bin
export ANDROID_STUDIO=$HOME/development/AndroidStudio
% grep SDK_ROOT .bashrc
export SDK_ROOT=$HOME/development/Android/SDK
export ANDROID_SDK_ROOT=$HOME/development/Android/SDK
Hope that answers some questions, sorry it took so long to give you a response.
I'm using CentOS 6. Since the CentOS repos have an older version of vim (7.2), I have compiled my own. However, I am having a weird issue. After running and exiting vim, I've found that my shell prompt has disappeared. Also, when I type, it does not appear in the terminal, nor does any output. What's weird is that it looks as though something is being outputted, since the cursor will move down the screen, but I can't see anything.
It must have something to do with compiling an X version, because it does not happen when I compile vim without X windows support (but I would like to have gvim, too). Here are the configuration options I supplied when compiling:
./configure --prefix=$OPT/Cellar/vim/7.4 --with-features=huge --enable-gui=auto --with-x --enable-xim
I also tried compiling without Xim. I've also tried explicitly setting --enable-gui=gtk2 (which is what gvim 7.2 from the CentOS repos uses).
I am using KDE4. The problem occurs both when running tcsh and bash. I have tried with $TERM set to both xterm and xterm-256color as well.
Additionally, running neither reset nor stty sane restores the shell to its proper behavior.
Any idea why my prompt disappears when running vim? As I missing a config flag or some other compilation issue?
I ultimately traced this down to an error in my ~/.vimrc file. I had some lines like this:
if has('gui_running')
# Gvim customization
endif
if has('gui_gtk2')
# GTK2-specific Gvim customization
endif
The second block was running every time vim loaded. Because it changed some window geometry, it ended up messing up bash; when I dropped back to bash, bash thought it only had ~53 columns to work with (and possibly some other things were messed up).
Solution: The second if block should be inside the first.
whenever I run eclipse from the shortcut I am unable to correctly build some of my projects because the PATH variable that I configured in .bashrc doesn't get used.
When I run eclipse from my terminal, I can build all my projects perfectly fine because it's running through the correct shell.
The problem is that I want to use the PATH variable from my .bashrc without permanently having a terminal open. I tried this before, but every day I accidentally close the terminal that's running eclipse by accident and lose all my unsaved code.
Can anyone help me?
Your tooling probably utilizes the embedded eclipse terminal. This terminal does not start providing your login/user shell. So you need to set the eclipse terminal in your Eclipse preferences to start as --login shell:
Go to:
Preferences -> Terminal -> Local Terminal
and set
"Arguments" to "--login"
restart Eclipse and your users $PATH should be used from now on.
Edit /usr/share/applications/eclipse.desktop with write privileges, i.e. sudo gedit /usr/share/applications/eclipse.desktop
Change the setting Exec=/usr/bin/eclipse to Exec=bash -ic "/usr/bin/eclipse" and save
The underlying issue is that .bashrc is not loaded in a non-interactive shell. When you start Eclipse normally clicking on its symbol, .bashrc quits early. This solution applies to all programs that are defined by a .desktop file.
In contrast, bash -i opens an interactive shell, -c "" runs a command in that shell.
I can think of two options for this problem:
write a small script, export those vars or source your .bashrc before you start your eclipse.
define those variables in /etc/environment. then they are not user-scope any more.
I prefer the 1st option.
Create simple script
#!/bin/bash
source /home/user/.environment_variables
/home/user/eclipse_cpp/eclipse -Duser.name="My Name"
2.
Next put your all system variables in file /home/user/.environment_variables (any file you want)
My looks like:
export COCOS_ROOT=/home/user/Projects/edukoala
export JAVA_HOME=/usr/lib/jvm/java-7-openjdk-amd64/
3.
Now you can delete your variables in .bashrc and put line
source /home/user/.environment_variables
Everything works fine :)
Well, this is already answered and the answer has been accepted. But this will also work for running your code using Eclipse. You can edit the Run Configurations and set the environment variable there. Then, Eclipse will pick up the variable from this setting while building.
I'm trying to configure vim as my primary coding program. I have figured how to compile single files, but when I go to execute the program from within vim, I keep getting a 127 error code. I have a aliased on my box to ./a.out, however when I issue the command :!a from vim, it doesn't work. :!./a.out does. Does anyone know why this is?
Aliases are defined in rc files that are sourced by interactive shell only and work only in interactive mode (vim does pass everything to shell, it never executes anything except the shell directly with fork+execve).
By default shell launched from vim starts in non-interactive mode hence bashrc is not read and no aliases are defined (though even if they were defined, they won’t be used in non-interactive mode). You may set
set shellcmdflag=-ic
, then shell will be launched in interactive mode and .bashrc file with your alias will be read.
Aliases are a feature of your shell (say, bash). ! operator directly executes a file - shell never sees it, and can't do alias expansion on it. (EDIT: See ZyX)
If you want to make executing your ./a.out easy, you can do something like:
command XX !./a.out
then you can do :XX. Or,
nnoremap X :!./a.out<CR>
then a single key X will suffice.
Another option (which is frowned upon for security reasons) is to add the current directory to your PATH:
PATH="./$PATH"
which will allow you to run your program with just a.out as opposed to ./a.out. If you then compile your program with (a language-dependent isomorphism of) -o switch, you can have it named just a:
gcc -o a foo.c
Given this, !a would work. Use at your own risk.
EDIT: ZyX is correct-er. :) I'll leave the answer here for the other information.