Re: typedef struct WindowClause misleading comments

David Rowley <dgrowleyml(at)gmail(dot)com> writes:
> On Wed, 10 Jun 2026 at 04:54, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> +1. There are also other comments about query jumbling in nodes/*.h that
>> seem pretty information-free now. They might have been helpful before
>> we invented query_jumble_ignore and related annotations, but now they
>> seem just duplicative.

> Here's a patch for that. I did leave a few comments which mention a
> reason why a particular field is ignored. That seems like it could be
> useful. I think I've got all the ones that just talk about what's
> included or ignored.

All these changes look good, but I have a few more suggestions,
attached as a delta on top of yours. Notably

* - query_jumble_location: Mark the field as a location to track. This is
- * only allowed for integer fields that include "location" in their name.
+ * only used for fields of type ParseLoc, which otherwise are not jumbled.

If you look at how gen_node_support.pl implements that annotation,
my revised statement is correct about the field type, and I don't
see anything that actually constrains the field name to be "location".
Maybe some earlier implementation behaved that way?

regards, tom lane