Re: Rewrite of zsh-newuser-install

[Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>]

I'll reply to this first and then look at the file itself when I have
more time ...

On Tue, Feb 9, 2021 at 3:15 PM dana <dana@xxxxxxx> wrote:
>
> > zmodload -F zsh/files b:zf_mkdir
>
> Then again, if you don't have the files module you probably don't have the
> newuser module either, so it's *probably* fine...?

This ends up being something for a package maintainer ... I think it's
OK to assume the existence of most of the standard modules, the
exceptions being e.g. zsh/cap, zsh/clone, zsh/pcre

> > PS1='%F{magenta}%$(( [#10] .2 * COLUMNS ))>..>%n%f@%F{cyan}%M%f%>>:'
> > PS1+='%B%F{blue}%$(( [#10] .2 * COLUMNS ))<..<%~%b%<< '

I'm one of those people who don't like PROMPT_SUBST.

> > PS1+='%(?,%F{green},%F{black}%K{red})%#%f%k '
>
> I like this, seems like a nice accessibility compromise

The black % on red background just disappears for me, it looks like
the prompt just ends in a red block.  %F{white} maybe?

> > RPS1=' %(1S,%S${SECONDS}s%s ,)%w %T'
>
> Again the time stamp actually feels *less* useful with transient RPROMPT

Also, you can get the elapsed time from something like ${${=$(fc -lD
-1)}[2]} instead of constantly resetting/reading $SECONDS (are the
fractional seconds really necessary?)

> > :zshrc:timer() { SECONDS=0 }

(see above)

> This function naming convention is OK with me.

I guess it's OK with me too.

> > # If $LS_COLORS hasn't been defined, try to generate it.
>
> Again, is this our business?
>
> And if it's *not* our business, i think we should take care to avoid making it
> appear set when it wasn't already.

Agree.

> (I'm not sure if the recent 'declared null'
> stuff impacts this [...])

Not in native zsh mode in any case.

> > setopt AUTO_CD ...
>
> I know some of these cd settings are OMZ defaults. I'm ambivalent about most
> of them, but i don't think i like CHASE_LINKS.

Agree about (not) CHASE_LINKS.

> > unsetopt CASE_GLOB
> > setopt EXTENDED_GLOB
>
> I like NO_CASE_GLOB.

I don't.  See the discussion leading up to my CASE_PATHS patch ...
NO_CASE_GLOB can cause globs that should succeed to fail, and globs
that should find a single result to return multiple results.

>I also personally like EXTENDED_GLOB, but i wonder if
> others will object?

I'm ambivalent.  I don't use it (have a wrapper function that enables
it when I need it, and hmm I just thought of what might be a better
way to do that) but it doesn't really cause that much issue if you
don't use the syntax.

> > setopt GLOB_DOTS
>
> Don't like that, seems unexpected and risky

Agree.

> > setopt GLOB_STAR_SHORT
> > setopt NUMERIC_GLOB_SORT
>
> Ambivalent

Me too.  I'm beginning to think I might like GLOB_STAR_SHORT but I
don't yet use it.

> > unsetopt CLOBBER
>
> Mixed feelings. I know a lot of power users like this, and i see the appeal,
> but it contradicts a lot of defaults and historical conventions

I personally do use NO_CLOBBER, but I abstain from the decision here.

> > unsetopt FLOW_CONTROL
> > setopt INTERACTIVE_COMMENTS
>
> I like these. Doubt the average user needs flow-control sequences, and
> interactive comments matches the default bash and OMZ behaviour

I like NO_FLOW_CONTROL, but I dislike INTERACTIVE_COMMENTS.  Not
strongly enough in the latter case to argue.