Re: [gnus git] branch master updated: =1= netrc.el: Return the value of the "default" entry.

Re: [gnus git] branch master updated: =1= netrc.el: Return the value of the "default" entry.

[Ted Zlatanov]

On Thu, 30 Sep 2010 22:07:19 +0200 Lars Magne Ingebrigtsen <larsi@quimbies.gnus.org> wrote: 

LMI> netrc.el: Return the value of the "default" entry.

What do you think about getting rid of netrc.el and moving that
functionality inside auth-source.el?  Otherwise we're basically
maintaining two notions of "default" and other concepts are duplicated
too.

Ted

Re: [gnus git] branch master updated: =1= netrc.el: Return the value of the "default" entry.

[Lars Magne Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> LMI> netrc.el: Return the value of the "default" entry.
>
> What do you think about getting rid of netrc.el and moving that
> functionality inside auth-source.el?  Otherwise we're basically
> maintaining two notions of "default" and other concepts are duplicated
> too.

Well, we don't want to get rid of ~/.authinfo(.gpg), do we?  So we need
to parse the file according to its syntax, and the manual page says that
"default" is valid...

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: [gnus git] branch master updated: =1= netrc.el: Return the value of the "default" entry.

[Ted Zlatanov]

On Thu, 30 Sep 2010 23:50:33 +0200 Lars Magne Ingebrigtsen <larsi@gnus.org> wrote: 

LMI> Ted Zlatanov <tzz@lifelogs.com> writes:
LMI> netrc.el: Return the value of the "default" entry.
>> 
>> What do you think about getting rid of netrc.el and moving that
>> functionality inside auth-source.el?  Otherwise we're basically
>> maintaining two notions of "default" and other concepts are duplicated
>> too.

LMI> Well, we don't want to get rid of ~/.authinfo(.gpg), do we?  So we need
LMI> to parse the file according to its syntax, and the manual page says that
LMI> "default" is valid...

I mean more from a code sprawl perspective.  It seems that the netrc.el
parser can comfortably live inside auth-source.el and then we can avoid
some of the query language ambiguity between the two APIs, plus it will
be easier to do delete and create operations.  It's really not a strong
feeling though, so feel free to say "no."

It would be good in the netrc.el header to provide examples of valid
netrc lines.  I have some in auth.texi I can add, do you want to give a
few as well?

Ted

Re: [gnus git] branch master updated: =1= netrc.el: Return the value of the "default" entry.

[Lars Magne Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> I mean more from a code sprawl perspective.  It seems that the netrc.el
> parser can comfortably live inside auth-source.el and then we can avoid
> some of the query language ambiguity between the two APIs, plus it will
> be easier to do delete and create operations.  It's really not a strong
> feeling though, so feel free to say "no."

No, it would make sense to only have the auth-source entry points.
Unless there are other third-party packages that use netrc-* stuff
directly... 

> It would be good in the netrc.el header to provide examples of valid
> netrc lines.  I have some in auth.texi I can add, do you want to give a
> few as well?

The examples in auth.texi are fine.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: [gnus git] branch master updated: =1= netrc.el: Return the value of the "default" entry.

[Ted Zlatanov]

On Fri, 01 Oct 2010 00:06:02 +0200 Lars Magne Ingebrigtsen <larsi@gnus.org> wrote: 

LMI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> I mean more from a code sprawl perspective.  It seems that the netrc.el
>> parser can comfortably live inside auth-source.el and then we can avoid
>> some of the query language ambiguity between the two APIs, plus it will
>> be easier to do delete and create operations.  It's really not a strong
>> feeling though, so feel free to say "no."

LMI> No, it would make sense to only have the auth-source entry points.
LMI> Unless there are other third-party packages that use netrc-* stuff
LMI> directly... 

I only see smtpmail.el inside Emacs.  Everyone else can simply adjust to
call auth-source-netrc-parse or whatever (if they want to bypass the
auth-sources mechanism).

Interestingly, Tramp has its own netrc parser.  Also ange-ftp.el has its
own.

Ted

Re: [gnus git] branch master updated: =1= netrc.el: Return the value of the "default" entry.

[Michael Albinus]

Ted Zlatanov <tzz@lifelogs.com> writes:

> Interestingly, Tramp has its own netrc parser.

Timing. IIRC, netrc.el didn't exist when I wrote this.

Compatibility with other (X)Emacs releases.

And it doesn't uses it for password detection, just for user/host names
exploring during minibuffer completion. No big deal.

> Also ange-ftp.el has its own.

Likely similar reasons.

> Ted

Best regards, Michael.

Re: [gnus git] branch master updated: =1= netrc.el: Return the value of the "default" entry.

[Lars Magne Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> I only see smtpmail.el inside Emacs.  Everyone else can simply adjust to
> call auth-source-netrc-parse or whatever (if they want to bypass the
> auth-sources mechanism).

Well, having a separate library that only does netrc parsing might be a
good idea, if other things need to parse those files, but without having
it be password-related.  So perhaps auth-sources should just use it as a
library. 

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen