From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
cc: Alexander Korotkov <a.korotkov@postgrespro.ru>,
        Andres Freund <andres@anarazel.de>,
        John Naylor <john.naylor@2ndquadrant.com>,
        Jeff Janes <jeff.janes@gmail.com>,
        Nikita Glukhov <n.gluhov@postgrespro.ru>,
        Tomas Vondra <tomas.vondra@2ndquadrant.com>,
        Oleg Bartunov <obartunov@postgrespro.ru>,
        Michael Paquier <michael@paquier.xyz>,
        Stas Kelvich <s.kelvich@postgrespro.ru>,
        PostgreSQL-development <pgsql-hackers@postgresql.org>,
        Thomas Munro <thomas.munro@enterprisedb.com>,
        David Steele <david@pgmasters.net>,
        Robert Haas <robertmhaas@gmail.com>,
        Pavel Stehule <pavel.stehule@gmail.com>
Subject: Re: jsonpath
In-reply-to: <dcff626f-d44d-ad96-ecc0-109601e2cfb9@2ndQuadrant.com>
References: 
 <CAMkU=1w1qBvoW82ZTFpAKae027R-2OHw-m6ALe0VQRNAFueBVA@mail.gmail.com>
 <CAPpHfdvbbVfgjm7Y0QQK8cmp4oXKQ1btXVk7arQKqHQqtX3TLg@mail.gmail.com>
 <CAPpHfdskvCZhGFHKZ49iX9=5z5Yk+2LXyKQ_XeCx7cHziV1znQ@mail.gmail.com>
 <CACPNZCt9dFP=LovsjXKNCtpEG32dkWLDVOo_9+XP406Ob_ptbw@mail.gmail.com>
 <CAPpHfdu6meyXZtyWL6vw0P54tJO8VCUL5my5qcfi29C4DDvACQ@mail.gmail.com>
 <20190324164458.srta7sz5ffizkqv5@alap3.anarazel.de>
 <CAPpHfdsK2+UsN9SCH0Ck8745S-eOeE5GEjzS0GoKwKYm=pt_Lw@mail.gmail.com>
 <CAPpHfduLp+WgnhgOWSebUL0UCw9uYrXAkZr2=9Js0eUe0f0wuA@mail.gmail.com>
 <6060.1553610709@sss.pgh.pa.us>
 <CAPpHfdvzvSSw6=C4JTDuxj1OjdNvVm7ncvLmY0fWr5dDH9qYfQ@mail.gmail.com>
 <CAPpHfdsXWZFWXuWXKKavtbw8taZ6QRzEwB6Nf2DX19z1HKFuPQ@mail.gmail.com>
 <16120.1553694533@sss.pgh.pa.us>
 <565a6051-b9df-3e94-b858-abce0f8b6436@2ndQuadrant.com>
 <CAPpHfdvkMdV_nQR3gO41GWJS=2Gjdrm6QhhYH2DwXJ+fMa6gBw@mail.gmail.com>
 <23e82578-b648-ab57-a4af-791abccff344@2ndQuadrant.com>
 <CAPpHfdsZvAQUQg6jjv4VU_Yn=_dhinmftZxLWK6EZuZY+Aq3Ng@mail.gm!
 ail.com> <dcff626f-d44d-ad96-ecc0-109601e2cfb9@2ndQuadrant.com>
Comments: In-reply-to Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
	message dated "Thu, 28 Mar 2019 09:10:51 -0400"
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-ID: <4970.1553779874.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Thu, 28 Mar 2019 09:31:14 -0400
Message-ID: <4971.1553779874@sss.pgh.pa.us>
Precedence: bulk

Andrew Dunstan <andrew.dunstan@2ndquadrant.com> writes:
> I mean repeated invocations of
> =C2=A0=C2=A0=C2=A0 psql -c "select '$ ? (@ like_regex \"pattern\" flag \=
"a\")'::jsonpath"
> These get crash, no crash, crash, no crash ...

That is just wacko ... but it does seem to support my thought of
a wild store somewhere.  The mechanism for this case might be
that memory layout is different depending on whether we had to
rebuild the relcache init file at session start or not.  Similarly,
the fact that the full test script reliably crashes might be
dependent on previous commands having left things in a certain
state.  Unfortunately that gets us little closer to a fix.

Has anybody gotten through a valgrind run on this code yet?

			regards, tom lane