Re: master d1c9310: DOn't use string-as-unibyte in Gnus

Re: master d1c9310: DOn't use string-as-unibyte in Gnus

[Katsumi Yamaoka]

Hi Lars,

On Tue, 31 Jan 2017 16:58:08 +0000, Lars Ingebrigtsen wrote:
> branch: master
> commit d1c931009004aef847105b7bac6b6ffafd985b82
> Author: Lars Ingebrigtsen <larsi@gnus.org>
> Commit: Lars Ingebrigtsen <larsi@gnus.org>

>     DOn't use string-as-unibyte in Gnus
[...]
>     * lisp/gnus/mml.el (mml-generate-mime-1): Ditto.
[...]
> --- a/lisp/gnus/mml.el
> +++ b/lisp/gnus/mml.el
> @@ -696,9 +696,10 @@ be \"related\" or \"alternate\"."
>  	      (set-buffer-multibyte nil)
>  	      (cond
>  	       ((cdr (assq 'buffer cont))
> -		(insert (string-as-unibyte
> +		(insert (encode-coding-string
>  			 (with-current-buffer (cdr (assq 'buffer cont))
> -			   (buffer-string)))))
> +			   (buffer-string))
> +			 'utf-8)))
>  	       ((and filename
>  		     (not (equal (cdr (assq 'nofile cont)) "yes")))
>  		(let ((coding-system-for-read mm-binary-coding-system))

This change breaks a jpeg image (at least).  Recipe:

・Visit an image file using `universal-coding-system-argument'
  with `binary' and `find-file', i.e.:
  C-x RET c binary RET C-x C-f FILENAME RET C-c C-c
・Open a message draft.
・Insert this mml tag:

<#part type="image/jpeg" disposition=inline buffer="IMAGE">

  Where "IMAGE" is the buffer name that visits the image file.
・Preview it by `C-c RET P' or send it by `C-c C-c'.

This means that a warning often seen recently

  ‘string-as-unibyte’ is an obsolete function (as of
      26.1); use ‘encode-coding-string’.

does not say the truth, does it?

Thanks.

Re: master d1c9310: DOn't use string-as-unibyte in Gnus

[Katsumi Yamaoka]

On Wed, 01 Feb 2017 19:00:51 +0900, Katsumi Yamaoka wrote:
> This means that a warning often seen recently

>   ‘string-as-unibyte’ is an obsolete function (as of
>       26.1); use ‘encode-coding-string’.

> does not say the truth, does it?

Sorry, I was wrong.  The problem seems to be due to encoding
a binary string.

Re: master d1c9310: DOn't use string-as-unibyte in Gnus

[Andreas Schwab]

On Feb 01 2017, Katsumi Yamaoka <yamaoka@jpl.org> wrote:

> Hi Lars,
>
> On Tue, 31 Jan 2017 16:58:08 +0000, Lars Ingebrigtsen wrote:
>> branch: master
>> commit d1c931009004aef847105b7bac6b6ffafd985b82
>> Author: Lars Ingebrigtsen <larsi@gnus.org>
>> Commit: Lars Ingebrigtsen <larsi@gnus.org>
>
>>     DOn't use string-as-unibyte in Gnus
> [...]
>>     * lisp/gnus/mml.el (mml-generate-mime-1): Ditto.
> [...]
>> --- a/lisp/gnus/mml.el
>> +++ b/lisp/gnus/mml.el
>> @@ -696,9 +696,10 @@ be \"related\" or \"alternate\"."
>>  	      (set-buffer-multibyte nil)
>>  	      (cond
>>  	       ((cdr (assq 'buffer cont))
>> -		(insert (string-as-unibyte
>> +		(insert (encode-coding-string
>>  			 (with-current-buffer (cdr (assq 'buffer cont))
>> -			   (buffer-string)))))
>> +			   (buffer-string))
>> +			 'utf-8)))

Shouldn't that be raw-text?

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: master d1c9310: DOn't use string-as-unibyte in Gnus

[Lars Ingebrigtsen]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> This change breaks a jpeg image (at least).  Recipe:

Hm; I'll revert the patch.  I think I need to consider each case more
closely than I did when I made the changes yesterday...

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

Re: master d1c9310: DOn't use string-as-unibyte in Gnus

[Andy Moreton]

On Wed 01 Feb 2017, Katsumi Yamaoka wrote:

> Hi Lars,
>
> On Tue, 31 Jan 2017 16:58:08 +0000, Lars Ingebrigtsen wrote:
>> branch: master
>> commit d1c931009004aef847105b7bac6b6ffafd985b82
>> Author: Lars Ingebrigtsen <larsi@gnus.org>
>> Commit: Lars Ingebrigtsen <larsi@gnus.org>
>
>>     DOn't use string-as-unibyte in Gnus
> [...]
>>     * lisp/gnus/mml.el (mml-generate-mime-1): Ditto.
> [...]
>> --- a/lisp/gnus/mml.el
>> +++ b/lisp/gnus/mml.el
>> @@ -696,9 +696,10 @@ be \"related\" or \"alternate\"."
>>  	      (set-buffer-multibyte nil)
>>  	      (cond
>>  	       ((cdr (assq 'buffer cont))
>> -		(insert (string-as-unibyte
>> +		(insert (encode-coding-string
>>  			 (with-current-buffer (cdr (assq 'buffer cont))
>> -			   (buffer-string)))))
>> +			   (buffer-string))
>> +			 'utf-8)))
>>  	       ((and filename
>>  		     (not (equal (cdr (assq 'nofile cont)) "yes")))
>>  		(let ((coding-system-for-read mm-binary-coding-system))
>
> This change breaks a jpeg image (at least).  Recipe:
>
> ・Visit an image file using `universal-coding-system-argument'
>   with `binary' and `find-file', i.e.:
>   C-x RET c binary RET C-x C-f FILENAME RET C-c C-c
> ・Open a message draft.
> ・Insert this mml tag:
>
> <#part type="image/jpeg" disposition=inline buffer="IMAGE">
>
>   Where "IMAGE" is the buffer name that visits the image file.
> ・Preview it by `C-c RET P' or send it by `C-c C-c'.
>
> This means that a warning often seen recently
>
>   ‘string-as-unibyte’ is an obsolete function (as of
>       26.1); use ‘encode-coding-string’.
>
> does not say the truth, does it?

I think that 'utf-8 (a variable length encoding) is wrong here, and that
the coding system should be 'binary. That seems to fix the image
corruption seen in the preview when following the recipe.

    AndyM

Re: master d1c9310: DOn't use string-as-unibyte in Gnus

[Eli Zaretskii]

> From: Andy Moreton <andrewjmoreton@gmail.com>
> Date: Wed, 01 Feb 2017 13:12:31 +0000
> 
> I think that 'utf-8 (a variable length encoding) is wrong here, and that
> the coding system should be 'binary. That seems to fix the image
> corruption seen in the preview when following the recipe.

I think Andreas is right, and raw-text is a better choice here.

Re: master d1c9310: DOn't use string-as-unibyte in Gnus

[Andy Moreton]

On Wed 01 Feb 2017, Eli Zaretskii wrote:

>> From: Andy Moreton <andrewjmoreton@gmail.com>
>> Date: Wed, 01 Feb 2017 13:12:31 +0000
>> 
>> I think that 'utf-8 (a variable length encoding) is wrong here, and that
>> the coding system should be 'binary. That seems to fix the image
>> corruption seen in the preview when following the recipe.
>
> I think Andreas is right, and raw-text is a better choice here.

Agreed - I read his reply just after posting.

    AndyM