Re: git: 47912ce2e613 - main - databases/sqlite3: update to 3.41.0

On 01/03/23 10:36, Florian Smeets wrote:
> On Tue, Feb 28, 2023 at 11:40:13PM +0100, Guido Falsi wrote:
>> On 27/02/23 23:49, Robert Clausecker wrote:
>>> The branch main has been updated by fuz:
>>>
>>> URL: https://cgit.FreeBSD.org/ports/commit/?id=47912ce2e613211aecf6ecdfc58681b7aca0df33
>>>
>>> commit 47912ce2e613211aecf6ecdfc58681b7aca0df33
>>> Author:     Pavel Volkov <pavelivolkov@gmail.com>
>>> AuthorDate: 2023-02-24 09:23:48 +0000
>>> Commit:     Robert Clausecker <fuz@FreeBSD.org>
>>> CommitDate: 2023-02-27 22:46:24 +0000
>>>
>>>       databases/sqlite3: update to 3.41.0
>>>       
>>>       Disable option DQS by default.
>>>       Add a note to UPDATING about this.
>>>       
>>
> 
> I'm answering as I'm the one that approved the commit.
> 
>> While I understand your motives this is a dangerous choice. There will
>> be a lot of runtime breakage in ports using sqlite3.
> 
> We are following upstream on this one, it's in the release notes.
> 
> https://www.sqlite.org/releaselog/3_41_0.html
> 
> 6.f.  The double-quoted string misfeature is now disabled by default for CLI
> builds. Legacy use cases can reenable the misfeature at run-time using the
> ".dbconfig dqs_dml on" and ".dbconfig dqs_ddl on" commands.
> 
> Accoring to https://www.sqlite.org/quirks.html#dblquote there has been a
> warning for 4 years warning about use of double quoted strings.
> 

I have no objection, as I stated I understand the situation, and don't 
disapprove of the change itself.

>>
>> I just discovered that deskutils/calibre in some parts of the code falls
>> for this.
>>
>> This is relatively easy to fix and I am testing patches. I'll also
>> submit it upstream:
>>
>> https://github.com/madpilot78/calibre/commit/e79cfdd3c6ab232fefa78dc185261ced7a209ae7
>>
>> but I'm sure a lot of ports will start acting up for no apparent reason
>> due to this.
> 
> Can someone back this up with some numbers? If I underestimated the
> consequences and this does indeed break more stuff we can/should revert,
> but we would deviate from upstream.

I did not mean to ask for a revert. I only wanted to cast a friendly 
warning that there will be some fallout.

I can't give you exact numbers, I have calibre failing (fix coming 
shortly), and have seen the bug report you also noticed (below)

> 
> I see we have a PR with at least two more broken ports 269889.


-- 
Guido Falsi <madpilot@FreeBSD.org>