Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Katsumi Yamaoka]

On Thu, 27 Apr 2017 17:38:27 -0400, Teodor Zlatanov wrote:
> diff --git a/lisp/auth-source.el b/lisp/auth-source.el
> index 7402ab2..5ad4277 100644
> --- a/lisp/auth-source.el
> +++ b/lisp/auth-source.el
[...]
> +        (return)))

This is a cl runtime function.  Could you replace it with something?
Thanks.

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Katsumi Yamaoka]

On Fri, 28 Apr 2017 16:53:35 +0900, Katsumi Yamaoka wrote:
> On Thu, 27 Apr 2017 17:38:27 -0400, Teodor Zlatanov wrote:
>> diff --git a/lisp/auth-source.el b/lisp/auth-source.el
>> index 7402ab2..5ad4277 100644
>> --- a/lisp/auth-source.el
>> +++ b/lisp/auth-source.el
> [...]
>> +        (return)))

> This is a cl runtime function.  Could you replace it with something?
> Thanks.

Sorry for not explaining enough.   That is an alias to the macro
`cl-return' but not expanded in the elc file.

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Ted Zlatanov]

On Fri, 28 Apr 2017 17:10:43 +0900 Katsumi Yamaoka <yamaoka@jpl.org> wrote: 

KY> On Fri, 28 Apr 2017 16:53:35 +0900, Katsumi Yamaoka wrote:
>> On Thu, 27 Apr 2017 17:38:27 -0400, Teodor Zlatanov wrote:
>>> diff --git a/lisp/auth-source.el b/lisp/auth-source.el
>>> index 7402ab2..5ad4277 100644
>>> --- a/lisp/auth-source.el
>>> +++ b/lisp/auth-source.el
>> [...]
>>> +        (return)))

>> This is a cl runtime function.  Could you replace it with something?

KY> Sorry for not explaining enough.   That is an alias to the macro
KY> `cl-return' but not expanded in the elc file.

Sorry, I don't know what's the proper fix. What breaks, and how should I
have caught this problem before it was comitted? Should I just replace
it with `cl-return' or is there more to do?

Thanks
Ted

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Stefan Monnier]

> Sorry, I don't know what's the proper fix. What breaks, and how should I
> have caught this problem before it was comitted?

"make" says:

      ELC      auth-source.elc
    
    In toplevel form:
    auth-source.el:782:1:Warning: Unused lexical variable ‘sym’
    auth-source.el:818:1:Warning: Unused lexical variable ‘sym’
    auth-source.el:1196:1:Warning: Unused lexical variable ‘ret’
    
    In end of data:
    auth-source.el:2080:1:Warning: the function ‘return’ is not known to be
        defined.

and while the 3 "unused ..." lines are due to my setting
byte-compile-force-lexical-warnings, the last warning should be present
for you as well.


        Stefan

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Glenn Morris]

Ted Zlatanov wrote:

> What breaks, and how should I have caught this problem before it was
> comitted?

You could have run "make check", to see a bunch of test failures due to
"return" being void.

I can't resist pointing out that this was flagged by
http://hydra.nixos.org/build/52055147

(BTW, your Mail-Followup-To header is broken: it only points to the list,
so dropping anyone you replied to who isn't subscribed.)

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Ted Zlatanov]

On Fri, 28 Apr 2017 12:36:40 -0400 Glenn Morris <rgm@gnu.org> wrote: 

GM> Ted Zlatanov wrote:
>> What breaks, and how should I have caught this problem before it was
>> comitted?

GM> You could have run "make check", to see a bunch of test failures due to
GM> "return" being void.

GM> I can't resist pointing out that this was flagged by
GM> http://hydra.nixos.org/build/52055147

Huh. Thanks for the pointer. I'm glad you didn't resist.

But I didn't see that failure on my system:

SUMMARY OF TEST RESULTS
-----------------------
Files examined: 154
Ran 2250 tests, 2233 results as expected, 17 skipped

So my "make check" doesn't flag that `return' as a void-function, and
the warning during compilation that Stefan pointed out is easy to miss.
Hydra may be doing something differently to catch this?

Stefan fixed it this morning by using `cl-return' and `cl-dolist'
instead, which is great, but I'd like to make sure I don't break the
build again. What am I missing in my test process?

Thanks
Ted

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Stefan Monnier]

> Stefan fixed it this morning by using `cl-return' and `cl-dolist'
> instead, which is great, but I'd like to make sure I don't break the
> build again. What am I missing in my test process?

My guess is that you byte-compiled the file from within your running
Emacs (which had `cl` loaded, so it was byte-compiled correctly) rather
than from `make`.


        Stefan

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Ted Zlatanov]

On Fri, 28 Apr 2017 14:32:29 -0400 Stefan Monnier <monnier@iro.umontreal.ca> wrote: 

>> Stefan fixed it this morning by using `cl-return' and `cl-dolist'
>> instead, which is great, but I'd like to make sure I don't break the
>> build again. What am I missing in my test process?

SM> My guess is that you byte-compiled the file from within your running
SM> Emacs (which had `cl` loaded, so it was byte-compiled correctly) rather
SM> than from `make`.

OK, I'll make sure to use a clean build next time. Thanks for the guidance.

Ted

Mail-Followup-To and Gmane and emacs-devel (was: master bd60ee2 1/5: auth-source: factor out parsers and add tests)

[Ted Zlatanov]

On Fri, 28 Apr 2017 12:36:40 -0400 Glenn Morris <rgm@gnu.org> wrote: 

GM> (BTW, your Mail-Followup-To header is broken: it only points to the list,
GM> so dropping anyone you replied to who isn't subscribed.)

I'm following up to the list since there are probably others using
gmane.emacs.devel instead of subscribing to the mailing list directly.
So whatever solution comes out of this discussion may help them too.

It's really annoying to get e-mail for discussions I see on
gmane.emacs.devel anyhow. This is the only way I've found to minimize
that e-mail.

So for a while now I've had it set this way. Actually, since I post
through Gmane, my headers are (the Gmane header is translated on the
Gmane side, Lars implemented that):

Mail-Copies-To: never
Gmane-Reply-To-List: yes

I'm not sure if there's a better approach, so I'll change my headers if
anyone has suggestions. But I really don't want the drive-by CC's back
in my mailbox.

Thanks
Ted

Re: master bd60ee2 1/5: auth-source: factor out parsers and add tests

[Stefan Monnier]

> so dropping anyone you replied to who isn't subscribed.)

Such as yours truly ;-)


        Stefan

Re: Mail-Followup-To and Gmane and emacs-devel

[Lars Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> So for a while now I've had it set this way. Actually, since I post
> through Gmane, my headers are (the Gmane header is translated on the
> Gmane side, Lars implemented that):
>
> Mail-Copies-To: never

This makes Message behave (i.e., don't include you in the Cc list)...

> Gmane-Reply-To-List: yes

This makes Gmane insert a Reply-To header in the mail, which is probably
what's making the problem here (i.e., dropping everybody from the Cc
list, even those who'd rather be on it).

> I'm not sure if there's a better approach, so I'll change my headers if
> anyone has suggestions. But I really don't want the drive-by CC's back
> in my mailbox.

I think the Gmane-Reply-To-List is the only realistic solution here for
this, but you could also just split to /dev/null all mail that has both
you and emacs-devel in the Cc headers.  

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Mail-Followup-To and Gmane and emacs-devel

[Ted Zlatanov]

On Mon, 08 May 2017 23:03:01 +0200 Lars Ingebrigtsen <larsi@gnus.org> wrote: 

LI> I think the Gmane-Reply-To-List is the only realistic solution here for
LI> this, but you could also just split to /dev/null all mail that has both
LI> you and emacs-devel in the Cc headers.  

Karl Fogel suggested letting Gnus deal with it, by suppressing duplicate
messages. Normally that only works in the same backend, but I could hook
into the gnus-registry and use it across backends. I don't know enough
of those internals, but the gnus-registry seems perfect for tracking by
message ID. Do you think that could work?

Ted

Re: Mail-Followup-To and Gmane and emacs-devel

[Lars Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> Karl Fogel suggested letting Gnus deal with it, by suppressing duplicate
> messages. Normally that only works in the same backend, but I could hook
> into the gnus-registry and use it across backends. I don't know enough
> of those internals, but the gnus-registry seems perfect for tracking by
> message ID. Do you think that could work?

Setting `gnus-suppress-duplicates' will mark articles with Message-IDs
you've seen already as read (by maintaining a cache of Message-IDs).  It
won't hide them completely, though, but that could be added...

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Mail-Followup-To and Gmane and emacs-devel

[Ted Zlatanov]

On Mon, 15 May 2017 22:07:17 +0200 Lars Ingebrigtsen <larsi@gnus.org> wrote: 

LI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> Karl Fogel suggested letting Gnus deal with it, by suppressing duplicate
>> messages. Normally that only works in the same backend, but I could hook
>> into the gnus-registry and use it across backends. I don't know enough
>> of those internals, but the gnus-registry seems perfect for tracking by
>> message ID. Do you think that could work?

LI> Setting `gnus-suppress-duplicates' will mark articles with Message-IDs
LI> you've seen already as read (by maintaining a cache of Message-IDs).  It
LI> won't hide them completely, though, but that could be added...

I can live with that, if it works across backends. I just looked and it
seems so; I didn't realize gnus-dup.el kept a global list. I read
gmane.emacs.devel most often, so I will have seen those message IDs.
I've now disabled `Gmane-Reply-To-List: yes` and let's see how it goes.

Thanks
Ted