Tarantool development patches archive
 help / color / mirror / Atom feed
From: Vladimir Davydov <vdavydov.dev@gmail.com>
To: Serge Petrenko <sergepetrenko@tarantool.org>
Cc: tarantool-patches@freelists.org, kostja@tarantool.org
Subject: Re: [PATCH v2 2/8] decimal: fix encoding numbers with positive exponent.
Date: Wed, 14 Aug 2019 14:56:43 +0300	[thread overview]
Message-ID: <20190814115643.GF13834@esperanza> (raw)
In-Reply-To: <53d58fa5cad01273d3e60fe034c8e74f023d3c0e.1565263272.git.sergepetrenko@tarantool.org>

On Thu, Aug 08, 2019 at 02:55:53PM +0300, Serge Petrenko wrote:
> When a number having a positive exponent is encoded, the internal
> decPackedFromNumber function returns a negative scale, which differs
> from the scale, returned by decimal_scale(). This leads to errors in
> decoding. Account for negative scale in decimal_pack() and
> decimal_unpack().
> 
> Follow-up #692
> ---
>  src/lib/core/decimal.c   | 10 +++++++---
>  test/unit/decimal.c      | 15 +++++++++++++--
>  test/unit/decimal.result | 11 ++++++++---
>  3 files changed, 28 insertions(+), 8 deletions(-)
> 
> diff --git a/src/lib/core/decimal.c b/src/lib/core/decimal.c
> index 6ef351f81..840aa5dfe 100644
> --- a/src/lib/core/decimal.c
> +++ b/src/lib/core/decimal.c
> @@ -33,6 +33,7 @@
>  #include "third_party/decNumber/decContext.h"
>  #include "third_party/decNumber/decPacked.h"
>  #include "lib/core/tt_static.h"
> +#include "lib/msgpuck/msgpuck.h"
>  #include <stddef.h>
>  #include <stdlib.h>
>  #include <float.h> /* DBL_DIG */
> @@ -340,12 +341,15 @@ char *
>  decimal_pack(char *data, const decimal_t *dec)
>  {
>  	uint32_t len = decimal_len(dec);
> -	*data++ = decimal_scale(dec);
> +	/* reserve space for resulting scale */
> +	char *svp = data++;
>  	len--;
>  	int32_t scale;
>  	char *tmp = (char *)decPackedFromNumber((uint8_t *)data, len, &scale, dec);
>  	assert(tmp == data);
> -	assert(scale == (int32_t)decimal_scale(dec));
> +	/* scale may be negative, when exponent is > 0 */
> +	assert(scale == (int32_t)decimal_scale(dec) || scale < 0);

This assertion now looks kinda lopsided/incomplete and therefore
confusing: we check positive scale but not negative scale. I'd rather
remove it altogether.

> +	mp_store_u8(svp, (int8_t)scale);

What is the reason to use mp_store/load helpers? AFAIU plain load/store
should work just fine (as they did before this patch).

Do I understand correctly that this particular patch must be committed
to all stable branches as well?

  parent reply	other threads:[~2019-08-14 11:56 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-08 11:55 [PATCH v2 0/8] Decimal indices Serge Petrenko
2019-08-08 11:55 ` [PATCH v2 1/8] lua: fix decimal comparison with nil Serge Petrenko
2019-08-12 21:16   ` Konstantin Osipov
2019-08-14 11:00   ` Vladimir Davydov
2019-08-14 22:17     ` Konstantin Osipov
2019-08-08 11:55 ` [PATCH v2 2/8] decimal: fix encoding numbers with positive exponent Serge Petrenko
2019-08-12 21:18   ` Konstantin Osipov
2019-08-13  9:00     ` [tarantool-patches] " Serge Petrenko
2019-08-14 22:21       ` Konstantin Osipov
2019-08-14 11:56   ` Vladimir Davydov [this message]
2019-08-08 11:55 ` [PATCH v2 3/8] lua/pickle: fix a typo Serge Petrenko
2019-08-12 21:18   ` Konstantin Osipov
2019-08-14 11:12   ` Vladimir Davydov
2019-08-14 11:15     ` Serge Petrenko
2019-08-08 11:55 ` [PATCH v2 4/8] lua: rework luaL_field types to support msgpack extensions Serge Petrenko
2019-08-12 21:23   ` Konstantin Osipov
2019-08-13 13:15     ` [tarantool-patches] " Serge Petrenko
2019-08-14 22:23       ` Konstantin Osipov
2019-08-15  8:27         ` Serge Petrenko
2019-08-16  8:06           ` Konstantin Osipov
2019-08-08 11:55 ` [PATCH v2 5/8] box: rework field_def and tuple_compare to work with mp_field_type instead of mp_type Serge Petrenko
2019-08-12 21:28   ` Konstantin Osipov
2019-08-08 11:55 ` [PATCH v2 6/8] decimal: allow to encode/decode decimals as MsgPack Serge Petrenko
2019-08-12 21:29   ` Konstantin Osipov
2019-08-12 21:34   ` Konstantin Osipov
2019-08-13 14:01     ` Serge Petrenko
2019-08-14 22:25       ` Konstantin Osipov
2019-08-08 11:55 ` [PATCH v2 7/8] decimal: add conversions to (u)int64_t Serge Petrenko
2019-08-12 21:39   ` Konstantin Osipov
2019-08-13 14:18     ` Serge Petrenko
2019-08-14 22:26       ` Konstantin Osipov
2019-08-14 22:29         ` Konstantin Osipov
2019-08-08 11:55 ` [PATCH v2 8/8] decimal: allow to index decimals Serge Petrenko
2019-08-08 13:42   ` Serge Petrenko
2019-08-12 21:41   ` Konstantin Osipov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190814115643.GF13834@esperanza \
    --to=vdavydov.dev@gmail.com \
    --cc=kostja@tarantool.org \
    --cc=sergepetrenko@tarantool.org \
    --cc=tarantool-patches@freelists.org \
    --subject='Re: [PATCH v2 2/8] decimal: fix encoding numbers with positive exponent.' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox