Re: Re: Toast,bytea, Text -blob all confusing

From: Hannu Krosing <hannu(at)tm(dot)ee>
To: Zeugswetter Andreas SB SD <ZeugswetterA(at)spardat(dot)at>
Cc: Joe Conway <joseph(dot)conway(at)home(dot)com>, Alex Pilosov <alex(at)acecape(dot)com>, Lincoln Yeoh <lyeoh(at)pop(dot)jaring(dot)my>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: Re: Toast,bytea, Text -blob all confusing
Date: 2001-08-30 14:32:18
Message-ID: 3B8E4E72.1566105A@tm.ee
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Zeugswetter Andreas SB SD wrote:
>
> > > > >For bytea, follow this rule: to escape a null character, use
> this:
> > > > >'\\0'. To escape a backslash, use this: '\\\\'.
>
> Can anybody explain in technical terms why this is implemented
> so inconveniently ?

I think that this has to to with making textin and textout behave
symmetrically, and the requirement that textout must produce a
valid C-string for ASCII transfer format.

> Since bytea is probably not very common among users yet
> we could imho still change it to not do double escapes.

But how ?

> Imho we need to decide where to do the escaping,
> eighter in the parser or in the input functions.

It would be probably hard to make the parser to _not_ unescape some
types, as it does not yet know it

> I think actually the backend parser has no business changing
> constants, he is imho only allowed to parse it, so he knows
> where a constant begins, and where it ends.

If it is any consolation then you have to write the inset of
a single \ from shell command so:

> psql -c "insert into t values('\\\\\\\\')"

;)

------------------
Hannu

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Thomas Lockhart 2001-08-30 14:53:25 Re: INTERVAL type: SQL92 implementation
Previous Message Peter Eisentraut 2001-08-30 13:28:10 Re: getting the oid for a new tuple in a BEFORE trigger