Re: pg_verifybackup: TAR format backup verification

From: Robert Haas <robertmhaas(at)gmail(dot)com>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Amul Sul <sulamul(at)gmail(dot)com>, Sravan Kumar <sravanvcybage(at)gmail(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: pg_verifybackup: TAR format backup verification
Date: 2024-10-01 14:39:51
Message-ID: CA+Tgmobazb7N0yURs1Cs0bmidFo+JTiqBFrZmw95B=bwPiW4dQ@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Mon, Sep 30, 2024 at 6:01 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > But then how do you think we should print
> > that? Cast to unsigned long long and use %llu?
>
> Our two standard solutions are to do that or to use UINT64_FORMAT.
> But UINT64_FORMAT is problematic in translatable strings because
> then the .po files would become platform-specific, so long long
> is what to use in that case. For a non-translated format string
> you can do either.

Here is an attempt at cleaning this up. I'm far from convinced that
it's fully correct; my local compiler (clang version 15.0.7) doesn't
seem fussed about conflating size_t with uint64, not even with -Wall
-Werror. I don't suppose you have a fussier compiler locally that you
can use to test this?

> > I don't understand what you think the widely-used, better solution is
> > here.
>
> What we just said above.

Respectfully, if you'd just said in your first email about this "I
understand that you were trying to be consistent with a format string
somewhere else, but I don't think that's a good reason to do it this
way, so please use %llu and insert a cast," I would have just said
"fine, no problem" and I wouldn't have been irritated at all. But you
seem determined to deny the existence of the argument that I made
instead of just disagreeing with it, and that's actually pretty
frustrating. I feel like you've wasted my time and your own to no
purpose, and made me feel stupid in the process, over something that
barely even matters. Anyone who has a control file bigger than 2GB has
... a lot of issues.

--
Robert Haas
EDB: http://www.enterprisedb.com

Attachment Content-Type Size
v1-0001-Try-to-use-uint64-to-size_t-for-file-sizes.patch application/octet-stream 9.3 KB

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2024-10-01 14:48:08 Re: pg_verifybackup: TAR format backup verification
Previous Message Laurenz Albe 2024-10-01 14:36:09 Re: First draft of PG 17 release notes