Re: [PATCH] Avoid redundant shift character in iconv output at block boundary (bug 17197)

[Florian Weimer <fweimer at redhat dot com>]

On 12/14/2015 05:23 PM, Martin Sebor wrote:
> On 12/14/2015 01:46 AM, Andreas Schwab wrote:
>> Martin Sebor <msebor@gmail.com> writes:
>>
>>> Great, thank you for confirming that.  Will you be committing
>>> the patch sometime soon or is something standing in the way?
>>
>> Lack of review.
> 
> Yes, I see you submitted the patch for review a number of times with
> no response.
> 
> For what it's worth, I've reviewed and successfully tested the patch.
> 
> I don't know the glibc commit process well enough to tell if someone
> else needs to review and approve it and who that might be in this
> case.  However, based on my reading of the Consensus page on the wiki
> I would expect anyone to be able to commit this patch with only no
> approval (I don't see the iconvdata component mentioned on in the
> Reviewers by component section on the MAINTAINERS page):
> 
>   Anyone can commit a locale related change where a bugzilla issue
>   exists, government sources are cited, common uses are cited, and
>   if there is an original author for the locale, that original author
>   ACKs the change. No developer review required. Post the patch and
>   ChangeLog to libc-alpha with a short message and then push the
>   commit.

But this is not a locale patch. :)

I think I understand what is going on here and how the patch fixes the bug.

Andreas, looking at the other direction (specifically, the first loop in
ibm930.c), it seems that an SI character while already in multi-byte
mode is illegal.  Does this mean the current encoder produces an illegal
output sequence?

Florian