Standardize library path

[maxux]

Configuration files are stored at ~/.config/jumpscale which is good standard.
But then you have ~/sandbox which is completely wrong.

Jumpscale library code, etc. should be installed at ~/.local/lib (and ~/.local/bin for binaries, if any).
This would completely standardize files path and looks more like real unix package :)

[xmonader]

I agree with that for sure, let's try to follow https://www.freedesktop.org/software/systemd/man/file-hierarchy.html as much as we can.

Here's the current implementation of j.dirs

class Dirs:

    HOMEDIR = os.path.expanduser("~")  # TODO: check homedir defined in sal.fs
    BASEDIR = os.path.join(HOMEDIR, "sandbox")
    BINDIR = os.path.join(BASEDIR, "bin")
    CFGDIR = os.path.join(BASEDIR, "cfg")  # TODO: check conflict with core.config_root..
    CODEDIR = os.path.join(BASEDIR, "code")
    VARDIR = os.path.join(BASEDIR, "var")
    LOGDIR = os.path.join(VARDIR, "log")
    JSCFGDIR = os.path.join(HOMEDIR, ".config/jumpscale")
    TEMPLATEDIR = os.path.join(VARDIR, "templates")
    TMPDIR = "/tmp/jumpscale"

I think when revisiting we need to consider the following

• we don't need TEMPLATEDIR anymore I think
• better we introduce CACHEDIR (for downloads we do instead of ~/sandbox/var/downloaded_packages/) or should be ~/.jumpscale/var/downloaded_packages
• ~/.local/lib/jumpscale ideally would have been used if we have some vendor data (right now that's completely managed by virtualenv, and when installing using pip it will use the /usr/lib or ~/.local/lib to install the deps
• CODEDIR by convention is ~/sandbox/code which we depend on in the git sal for cloning repos and that we can't have an a hidden directory, so we either introduce ~/code or ~/sandbox/code to the user filesystem, but if we are to introduce directories, then maybe ~/sandbox would serve us better as we can use it for transient configs, logs, .. etc as well
• apps configurations (e.g nginx, redis, .. etc) I think they should go under some directory maybe ~/.config/jumpscale/etc, but here it raises a problem what if we want to discard that dir for wrong configurations or regenerate everytime e.g in case of something dynamic like nginx? this directory ~/.config/jumpscale is backed up by convention and if something wrong you won't be able to have a successful restore
• Some apps e.g zoom, komodo, thunderbird some their config/logs in $HOME/.$appname, so maybe ~/.jumpscale/{etc,cfg,var,logs,cache} can be used for logs or then maybe it's okay to have sandbox dir~/sandbox/{etc,cfg,var,logs,cache}, and transient configs, logs and caching?

➜  ~ ls ~/.zoom 
data  im  logs  reports  screenCapture  shareItems
➜  ~ ls .thunderbird 
'Crash Reports'   i676rnyr.default  'Pending Pings'   profiles.ini
➜  ~ ls ~/.kiki 
kikicfg.py
➜  ~ ls ~/.komodoide 
12.0
➜  ~ ls ~/.komodoide/12.0/
autosave            codeintel3.log  history.sqlite.bak    prefs.xmlc    samples          tools
breakpoints.pickle  colors.less     interfaceChrome.less  pystderr.log  schemes          userChrome.less
classic.less        doc-state.xmlc  mutex.lock            pystdout.log  startup-env.tmp  view-state.xmlc
codeintel3.db       history.sqlite  prefs.xml             running.lock  toolbox.sqlite   XRE

if we are going to create ~/.jumpscale directory, wouldn't that be confused with configs of ~/.config/jumpscale or we merge both in ~/.jumpscale. What do you think @maxux?