Thread

Re: amcheck: support for GiST

Paul A Jungwirth <pj@illuminatedcomputing.com> — 2025-12-16T15:24:33Z

Hello,

On Wed, Oct 22, 2025 at 11:58 AM Kirill Reshke <reshkekirill@gmail.com> wrote:
>
> > 1) There are several typos in verify_gist.c:
> >
> >  downlinks -> downlink (header comment)
> >  discrepencies -> discrepancies
> >  Correctess -> Correctness
> >  hande -> handle
> >  Initaliaze -> Initialize
> >  numbmer -> number
> >  replcaed -> replaced
> >  aquire -> aqcuire
> >
> > 2) Copyright year is 2023 in the patch. Time flies:)
>
> These two are (trivially) fixed.

I found a few more typos. Maybe one is left over from Arseniy's
review. Referencing the latest patch files from Andrey, here is what I
see:

in v20251216-0002-Add-gist_index_check-function-to-verify-Gi.patch:

> This function traverses GiST with a depth-fisrt search and checks

"fisrt" should be "first".

> This traverse takes lock of any page until some discapency found.

"discapency" should be "discrepancy"

> To re-check suspicious pair of parent and child tuples it aqcuires

"aqcuires" should be "acquires"

amcheck.sgml:

+      require tuple adjustement) and page graph respects balanced-tree

"adjustement" should be "adjustment"

Also the Makefile ordering is not quite right:

--- a/contrib/amcheck/Makefile
+++ b/contrib/amcheck/Makefile
@@ -4,16 +4,17 @@ MODULE_big    = amcheck
 OBJS = \
    $(WIN32RES) \
    verify_common.o \
+   verify_gist.o \
    verify_gin.o \
    verify_heapam.o \
    verify_nbtree.o

We should put verify_gist.o after verify_gin.o.

Yours,

-- 
Paul              ~{:-)
pj@illuminatedcomputing.com