On Wed, Sep 26, 2012 at 01:54:51PM +0200, intrigeri wrote:
> >> +++ b/config/chroot_local-includes/usr/local/sbin/htpdate
> >> + [ 'proxy|p:s', "what to pass to curl's --socks5-hostname" ],
>
> > According to its manpage, curl will use the following environment
> > variable: `http_proxy`, `HTTPS_PROXY`, `FTP_PROXY`, `ALL_PROXY`.
> > I wonder if it would not make the behaviour of htpdate more
> > consistent to manually unset those variables before running `curl`.
> > Otherwise, htpdate could use a proxy with '-p' specified on the
> > command-line.
>
> Consistent with what? All this paragraph of yours is very unclear to
> me, and I'm not sure what you mean. Please rephrase, e.g. point out
> a specific potential problem you are envisioning :)
htpdate lists a "--proxy" option. I may assume that when I don't specify
this option, it will not use a proxy at all. But, the current code will
still use a proxy if HTTPS_PROXY or ALL_PROXY are set. I think this
is confusing.
> > I have a tickling feeling that this list will get outdated...
>
> Are talking of the list (of links to configuration files) that follows
> the introduction sentence you are quoting, or what?
>
> If you are, well, right, but this is a general problem with our entire
> design document. That would be partly addressed by a check for broken
> links, and additional strictness on the design doc front when
> reviewing/merging branch.
Yes, I was talking of the links to the configuration files. But yes, I
don't have a better idea.
> > Uh... and actually, those changes might require to add some more
> > tests to the checklist. What do you think?
>
> I'll think of it later today or tomorrow.
Neat! :)
--
Ague
