Re: pure parsers and reentrant scanners

On 27.12.24 10:19, Heikki Linnakangas wrote:
> On 26/12/2024 20:27, Peter Eisentraut wrote:
>> On 22.12.24 22:43, Andreas Karlsson wrote:
>>> On 12/19/24 9:57 PM, Peter Eisentraut wrote:
>>>> Here is an updated patch set on top of what has been committed so
>>>> far, with all the issues you pointed out addressed.
>>>
>>> Other than the discussion of how old versions of flex we should
>>> support I think this set of patches is ready to be committed. I
>>> looked at it again and everything looks good.
>>
>> I have committed these except the plpgsql one, which was still work in
>> progress.  But I have progressed on this now and also converted the
>> parser and put the local state into yyextra.  This gets rid of all
>> internal global state now.  The patches for this are attached.  It's a
>> lot of churn, but otherwise pretty standard stuff.
>
> Looks good to me.
>
>> Along the way I noticed that the flex documentation now recommends a
>> different way to set the yyextra type.  So I have changed the ones we
>> already have to that newer style.  I inspected the generated C code
>> and there wasn't any significant difference, so I'm not sure, but I
>> figure if we're making changes in this area we might as well use the
>> modern style.
>
> +1. According to the flex NEWS file, this syntax was added in flex
> 2.5.34, and we already require 2.5.35.

These have been committed. This concludes the main body of this work.

I'll let it take a lap around the buildfarm and will follow up in a few
days on what if anything lapwing has to say about it in terms of
warnings and what we want to do about it.