[systemd-devel] [PATCH 1/9] util: Add unescape_first_word()

[Lennart Poettering]

On Fri, 19.06.15 10:56, Richard Maw (richard.maw at codethink.co.uk) wrote:

> On Thu, Jun 18, 2015 at 08:30:22PM +0200, Lennart Poettering wrote:
> > On Thu, 28.05.15 13:02, Richard Maw (richard.maw at codethink.co.uk) wrote:
> > 
> > > This is a superset of the functionality of unquote_first_word, allowing
> > > non-whitespace separators, and doesn't interpret quotes unless
> > > UNQUOTE_QUOTES is included in flags.
> > 
> > Hmm, makes sense, but I'd actually just have one function
> > extract_first_word() then, which replaces unquote_first_word() but has
> > the signature of your unescape_first_word(). It would take the
> > separators parameter, which would default to WHITESPACE if passed as
> > NULL. THe flags should all be renamed EXTRACT_xyz instead of
> > UNQUOTE_xyz then, and EXTRACT_UNQUOTE should be a prominent flag.
> 
> Thanks, I was sweating the nomeclature and wasn't happy with what I came up
> with, but couldn't think of anything better.
> 
> > Then, all our current users of unquote_first_word() should be changed
> > to use this new call.
> > 
> > Does that make sense?
> 
> Sure, I was just a little wary of making such wide changes across
> the codebase.

We are not afraid of refactoring things like this, if it makes our
internal APIs simpler and cleaner!

> I also saw a couple of TODOs to convert uses of FOREACH_WORD family to the
> unquote_many_words family. I'll see if I can feasibly convert those while I'm
> changing string handling elsewhere.

Yes, we should stop using FOREACH_WORD, and move everything to
extract_first_word() loops, but I fear that's a major
endeavour... Would love patches for that, though.

Moving things over would slightly change behaviour of systemd when we
parse things, but I am sure it's a good thing actually. 

Lennart

-- 
Lennart Poettering, Red Hat