# [PATCH] number_format enhancements

- Thread: https://externals.io/message/24667
- Original archive: https://news-web.php.net/php.internals/24667
- Started by Matt W on 2006-07-19T10:54:34+00:00
- Messages: 8
- Reply by email to internals@lists.php.net
- More for agents: search with https://externals.io/search.md?q={query} · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt

---

## #24667 — Matt W — 2006-07-19T10:54:34+00:00

URL: https://externals.io/message/24667#24667

Hi,

There are a couple things I don't like about how `number_format()` works
currently, so I'm hoping the functionality changes this patch provides can
be added. :-)

First is that when you're working with whole numbers and/or don't want any
decimal places, both decimal-related parameters are needed before the
thousands separator.  I always thought it would be logical to pass the
thousands separator as param #2 in those cases.  That's what my changes
allow if it's a non-numeric string.

Second is that when decimals places ARE specified, I would like to be able
to have trailing zeros/point removed.  (Like "(string) round(...)" but never
scientific notation.)  I'm sure I'm not the only one that has wanted grouped
thousands and/or custom decimal point *without* trailing 0's. :-)  The patch
allows this by specifying negative decimal places -- e.g. -2 is 2 decimals
without extra zeros.

Are these enhancements all right?  If they're agreed to, I'll create more
 `number_format()` tests right away if desired.  Just wanted to get any
thoughts on the patch first.

Also, I changed PHP_ROUND_WITH_FUZZ to skip unnecessary steps if places==0
(round() is nearly 20% faster then, in my quick tests), corrected the
misspelling of "separator" in the proto, and simplified some of the code.

http://realplain.com/php/number_format_enhancements.diff
http://realplain.com/php/number_format_enhancements_5_2.diff


Thanks,
Matt

---

## #24875 — Matt W — 2006-07-22T11:28:27+00:00

URL: https://externals.io/message/24667#24875 · In reply to: #24667

Hi,

Haven't heard any opinions on these functionality changes...  (Though I know
more significant things keep you busy!)

Do many other people use `number_format()`??  Wouldn't allowing thousands
separator as param #2 simplify things?

Any issues with the semantics of operation?  Negative precision to specify
trailing zero removal seems fine at least, even though that's different than
 `round()`'s behavior.

I've now made some tests, including changes of course, in case you were just
waiting for them?  Hope they're enough.

http://realplain.com/php/number_format.phpt


Thanks,
Matt


----- Original Message -----
From: "Matt W"
Sent: Wednesday, July 19, 2006


> Hi,
>
> There are a couple things I don't like about how `number_format()` works
> currently, so I'm hoping the functionality changes this patch provides can
> be added. :-)
>
> First is that when you're working with whole numbers and/or don't want any
> decimal places, both decimal-related parameters are needed before the
> thousands separator.  I always thought it would be logical to pass the
> thousands separator as param #2 in those cases.  That's what my changes
> allow if it's a non-numeric string.
>
> Second is that when decimals places ARE specified, I would like to be able
> to have trailing zeros/point removed.  (Like "(string) round(...)" but
never
> scientific notation.)  I'm sure I'm not the only one that has wanted
grouped
> thousands and/or custom decimal point *without* trailing 0's. :-)  The
patch
> allows this by specifying negative decimal places -- e.g. -2 is 2 decimals
> without extra zeros.
>
> Are these enhancements all right?  If they're agreed to, I'll create more
> `number_format()` tests right away if desired.  Just wanted to get any
> thoughts on the patch first.
>
> Also, I changed PHP_ROUND_WITH_FUZZ to skip unnecessary steps if places==0
> (round() is nearly 20% faster then, in my quick tests), corrected the
> misspelling of "separator" in the proto, and simplified some of the code.
>
> http://realplain.com/php/number_format_enhancements.diff
> http://realplain.com/php/number_format_enhancements_5_2.diff
>
>
> Thanks,
> Matt

---

## #24885 — bertrand Gugger — 2006-07-22T19:47:03+00:00

URL: https://externals.io/message/24667#24885 · In reply to: #24875

Bonsoir Matt,
Matt W wrote:
> Hi,
> 
> Haven't heard any opinions on these functionality changes...  (Though I know
> more significant things keep you busy!)
> 
> Do many other people use `number_format()`??  Wouldn't allowing thousands
> separator as param #2 simplify things?
> 
> Any issues with the semantics of operation?  Negative precision to specify
> trailing zero removal seems fine at least, even though that's different than
> `round()`'s behavior.
> 
> I've now made some tests, including changes of course, in case you were just
> waiting for them?  Hope they're enough.
> 
> http://realplain.com/php/number_format.phpt
I did not look into that .phpt link
I belong only to userland , so have less to tell here.

Give some basic userland examples about what you propose to change, 
please ...
(I guess these are a part of the .phpt)

Bravo for the way you propose things.

Kind regards,
--
toggg

> ----- Original Message -----
> From: "Matt W"
> Sent: Wednesday, July 19, 2006
> 
> 
> 
> > […]
> 
> never
> 
> > […]
> 
> grouped
> 
> > […]
> 
> patch
> 
> > […]

---

## #24891 — bertrand Gugger — 2006-07-22T23:21:07+00:00

URL: https://externals.io/message/24667#24891 · In reply to: #24875

Matt W wrote:
> Hi,
> 
> Haven't heard any opinions on these functionality changes...  (Though I know
> more significant things keep you busy!)
> 
> Do many other people use `number_format()`??  Wouldn't allowing thousands
> separator as param #2 simplify things?
> 
> Any issues with the semantics of operation?  Negative precision to specify
> trailing zero removal seems fine at least, even though that's different than
> `round()`'s behavior.
> 
> I've now made some tests, including changes of course, in case you were just
> waiting for them?  Hope they're enough.
> 
> http://realplain.com/php/number_format.phpt
Sorry , I did not come better than the attached test
copies http://toggg.info/number_format.phpt & 
http://toggg.info/number_format.diff
(you don't need "\n")
I wonder why lines are fliping , I guess it's a diff effect

Let me know how I apply your patch , so I get it blank :)
-- 
toggg

---

## #24901 — Matt W — 2006-07-23T10:22:17+00:00

URL: https://externals.io/message/24667#24901 · In reply to: #24885

Hi,

----- Original Message -----
From: "bertrand Gugger"
Sent: Saturday, July 22, 2006


> Bonsoir Matt,
> Matt W wrote:
> > […]
know
> > […]
specify
> > […]
than
> > […]
just
> > […]
> I did not look into that .phpt link
> I belong only to userland , so have less to tell here.
>
> Give some basic userland examples about what you propose to change,
> please ...
> (I guess these are a part of the .phpt)
>
> Bravo for the way you propose things.
>
> Kind regards,
> --
> toggg

In the .phpt file, it is userland code where you can see examples of how the
changes work. :-)

I'm on Windows, so I don't know how you apply the .diff to your source
download -- shouldn't diff work somehow? :-/  Well, I uploaded the full
math.c file for 5.2 and 6 if you want to grab that to replace yours
(ext/standard/math.c) and experiment with the changes.

http://realplain.com/php/math.c-number_format_enhancements.tar.gz (5.2/6.0
directories with the file.)

Glad you like the idea. :-)  I just hope functionality like this can be
added -- using my patch or not, or slightly different semantics...  All I
want is a no extra 0's option and not be forced to include 2 unnecessary
decimal-related parameters to "get to" thousands separator when I don't want
any decimals! ;-D


Matt

---

## #24917 — bertrand Gugger — 2006-07-24T04:30:28+00:00

URL: https://externals.io/message/24667#24917 · In reply to: #24901

Matt W wrote:
> Hi,
> > […]
> 
> In the .phpt file, it is userland code where you can see examples of how the
> changes work. :-)
Sure, but I was meaning , in so far you extend the functionality of 
number_format , then you will need to extend the doc.
Some draft of this doc extension would be usefull , even in this test 
phase and especially for "userland" people.
> 
> I'm on Windows, so I don't know how you apply the .diff to your source
> download -- shouldn't diff work somehow? :-/  Well, I uploaded the full
> math.c file for 5.2 and 6 if you want to grab that to replace yours
> (ext/standard/math.c) and experiment with the changes.
> 
> http://realplain.com/php/math.c-number_format_enhancements.tar.gz (5.2/6.0
> directories with the file.)
I applied the patch 
http://realplain.com/php/number_format_enhancements_5_2.diff to the snap:
[php5.2-200607222030]$ patch -p3 -b -i 
/home/bertrand/A/mattW/number_format/5.2/number_format_enhancements_5_2.diff

No problem , I could check I get the right ext/standard/math.c by 
diffing whith the math.c you published in this tar.gz.
$ diff ext/standard/math.c ~/A/mattW/number_format/5.2/
981c981
<       int is_negative=0, trim_dec=0;
---
 >       int is_negative = 0, trim_dec = 0;
so just a litle cleaning difference.
As a note , the patch also corrected ext/standard/basic_functions.c 
(seperator ==> separator) which I guess is needed too :)

make clean , configure , make

Then , using your original .phpt :
$ export 
TEST_PHP_EXECUTABLE=/home/bertrand/php/php5.2-200607222030/sapi/cli/php
$ sapi/cli/php run-tests.php ~/A/mattW/number_format.phpt

=====================================================================
CWD         : /home/bertrand/php/php5.2-200607222030
PHP         : /home/bertrand/php/php5.2-200607222030/sapi/cli/php
 `PHP_SAPI`    : cli
 `PHP_VERSION` : 5.2.0-dev
ZEND_VERSION: 2.2.0
 `PHP_OS`      : Linux - Linux ancilla.toggg.net 2.6.12-1.1381_FC3 #1 Fri 
Oct 21 03:46:55 EDT 2005 i686
INI actual  : /usr/local/lib/php.ini
More .INIs  :
Extra dirs  :
=====================================================================
Running selected tests.
PASS `number_format()` function [/home/bertrand/A/mattW/number_format.phpt]

Nice !
Just not sure if the .phpt wouldn't need some more cases ("classical" ones).
> 
> Glad you like the idea. :-)  I just hope functionality like this can be
> added -- using my patch or not, or slightly different semantics...  All I
> want is a no extra 0's option and not be forced to include 2 unnecessary
> decimal-related parameters to "get to" thousands separator when I don't want
> any decimals! ;-D
Actually , I was talking about the way you *propose* things , not about 
the things themselves.

Errmmm.... not sure I'm not boring the list with such "userland" blabla,
I just hope it can serve some of my "ignorant" colleagues.

Cheers
-- 
toggg

---

## #24922 — Matt W — 2006-07-24T08:34:05+00:00

URL: https://externals.io/message/24667#24922 · In reply to: #24917

Hi bertrand,

----- Original Message ----- 
From: "bertrand Gugger"
Sent: Sunday, July 23, 2006

> Matt W wrote:
> > […]
the
> > […]
> Sure, but I was meaning , in so far you extend the functionality of
> number_format , then you will need to extend the doc.
> Some draft of this doc extension would be usefull , even in this test
> phase and especially for "userland" people.

Oh, I see.  I don't know anything about changing docs themselves (as far as
patch-type stuff goes). :-)  Thought that was handled by someone else...
Are you one of "the PHP people?"

I *could* do a draft of doc changes, of course, *if* I knew the changes
would be applied.  But I've not heard from anyone but you; maybe since it's
just a "little ol' function." :-/

> > […]
(5.2/6.0
> > […]
> I applied the patch
> http://realplain.com/php/number_format_enhancements_5_2.diff to the snap:
> [php5.2-200607222030]$ patch -p3 -b -i
>
/home/bertrand/A/mattW/number_format/5.2/number_format_enhancements_5_2.diff
>
> No problem , I could check I get the right ext/standard/math.c by
> diffing whith the math.c you published in this tar.gz.
> $ diff ext/standard/math.c ~/A/mattW/number_format/5.2/
> 981c981
> <       int is_negative=0, trim_dec=0;
> ---
> > […]
> so just a litle cleaning difference.
> As a note , the patch also corrected ext/standard/basic_functions.c
> (seperator ==> separator) which I guess is needed too :)

Oops, yeah, you found the whitespace changes I made in the patch. :-P

Well, basic_functions.c typo fix isn't needed for the functionality changes,
of course. :-)

> make clean , configure , make
>
> Then , using your original .phpt :
> $ export
> TEST_PHP_EXECUTABLE=/home/bertrand/php/php5.2-200607222030/sapi/cli/php
> $ sapi/cli/php run-tests.php ~/A/mattW/number_format.phpt
>
> =====================================================================
> CWD         : /home/bertrand/php/php5.2-200607222030
> PHP         : /home/bertrand/php/php5.2-200607222030/sapi/cli/php
> `PHP_SAPI`    : cli
> `PHP_VERSION` : 5.2.0-dev
> ZEND_VERSION: 2.2.0
> `PHP_OS`      : Linux - Linux ancilla.toggg.net 2.6.12-1.1381_FC3 #1 Fri
> Oct 21 03:46:55 EDT 2005 i686
> INI actual  : /usr/local/lib/php.ini
> More .INIs  :
> Extra dirs  :
> =====================================================================
> Running selected tests.
> PASS `number_format()` function [/home/bertrand/A/mattW/number_format.phpt]
>
> Nice !
> Just not sure if the .phpt wouldn't need some more cases ("classical"
ones).

Yeah, I don't know how many variations are desired...

> > […]
I
> > […]
want
> > […]
> Actually , I was talking about the way you *propose* things , not about
> the things themselves.

Ooooh! Hehe  Thanks then. :-)

> Errmmm.... not sure I'm not boring the list with such "userland" blabla,
> I just hope it can serve some of my "ignorant" colleagues.
>
> Cheers
> -- 
> toggg


Matt

---

## #24925 — bertrand Gugger — 2006-07-24T09:06:12+00:00

URL: https://externals.io/message/24667#24925 · In reply to: #24922

Matt W wrote:
> > […]
> 
> 
> Oh, I see.  I don't know anything about changing docs themselves (as far as
> patch-type stuff goes). :-)
Yes , it's heavy docbook thing , so you need to be a freak yourself :)
> Thought that was handled by someone else...
Sure , there's a great team of people there , translators , with their 
own list , etc.
> Are you one of "the PHP people?"
I belong only to one of the first "userland" layer , pear , though I'm 
currently inactive there.
> 
> I *could* do a draft of doc changes, of course, *if* I knew the changes
> would be applied.  But I've not heard from anyone but you; maybe since it's
> just a "little ol' function." :-/
Yes , case it's accepted , just a .txt (I mean no docbook) with the 
needed additions to the manual page,
sent or bugged to the doc team would be helpfull.
-- 
toggg