X-Authentication-Warning: delorie.com: mail set sender to djgpp-workers-bounces using -f
X-Recipient: djgpp-workers AT delorie DOT com
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=gmail.com; s=20120113;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type;
        bh=a4DR/ikRc5C6voOSRYVklA1Z+CfWstTtP+9lSnjAmr8=;
        b=y3f8RQFGuVa6NZvJyNevI/kgCNdWETdeEtYgiLjan3SRf2hrBIr9gU/8wBlvWnumBb
         N5zeukMr7grVyCXdUpunaLO8DHO5YexDvRCutbSCRazkC9nl+g9GoJHJyrQeLPft4TkK
         GOLgSqPSUd3AkG3lA71vqyX9hxxWI10Oj0z4rFYhhpRx8zrsJTKF7Wb77vCtbIBgZqg6
         X6arBl2nuXssV5Ou+xndZ9+mFvh38TrhCGLA2bJ4EtAW2H50qd31Kr/UDC26VUDp1zTa
         iODJDdadSM3QNxo5/OYl7GpY7S/1lC6DHL3SIv305JnodZ10HInN/MO869+OMcETpAt1
         w2KA==
MIME-Version: 1.0
X-Received: by 10.42.20.6 with SMTP id e6mr31606791icb.29.1398118986605; Mon,
 21 Apr 2014 15:23:06 -0700 (PDT)
In-Reply-To: <CAA-ihx9Je2xXChvz1+3_dSLt+KbrkzsymsbXFH=_5wqOqXzCMQ@mail.gmail.com>
References: <CAA2C=vBamXSU9Eh53WTLos=nemZs+uJ09QPF9Y8Jj4OSdZPsyQ AT mail DOT gmail DOT com>
	<53555BFD DOT 4000202 AT gmx DOT de>
	<CAA-ihx9Je2xXChvz1+3_dSLt+KbrkzsymsbXFH=_5wqOqXzCMQ AT mail DOT gmail DOT com>
Date: Tue, 22 Apr 2014 01:23:06 +0300
Message-ID: <CAA2C=vBHY=EYBWt+CDFL+su=vsy4yhG4HDvwpnB6i1qDDwDqzw@mail.gmail.com>
Subject: Re: ctime.c changes add about 4.5k more size
From: Ozkan Sezer <sezeroz AT gmail DOT com>
To: djgpp-workers AT delorie DOT com
Content-Type: text/plain; charset=UTF-8
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

On 4/22/14, Rugxulo <rugxulo AT gmail DOT com> wrote:
> Hi,
>
> On Mon, Apr 21, 2014 at 12:57 PM, Juan Manuel Guerrero
> <juan DOT guerrero AT gmx DOT de> wrote:
>> Am 21.04.2014 08:44, schrieb Ozkan Sezer:
>>
>>> Recent(ish) src/libc/ansi/time/ctime.c changes add about 4.5k more
>>> size to the final stripped binary:
>>>
>>> #include <time.h>
>>> int main () {
>>>   return time(NULL);
>>> }
>>>
>>> Linking against v2.04 from 2011-10-01 gives a 91648 bytes a.exe,
>>> whereas linking against v2.04 from 2014-04-20 gives a 96256 bytes
>>> a.exe. (not mentioning at all the crazy sizes themselves which is
>>> irrelevant to the present case at hand.)  A "return 0" instead of
>>> a "return time(NULL)" yields a 52736 bytes exe, so ctime.c stuff
>>> is adding about 43k size.
>>>
>>> time() calls gettimeofday() which calls 0x2c and 0x2a dos functions
>>> and calls ctime.c::mktime() where the additional bloat happens.
>>>
>>> Is there no other way of reducing the code size here?
>
> Not automatically since COFF (nor PE/COFF) doesn't have proper support
> for --gc-sections in GNU ld, last I heard.
>
>> The time zone database  format has changed so that the original ctime
>> code
>> of djgpp (code before 2011-10-01) had to be replaced (aka ported) by the
>> ctime code provided by time zone source code.
>>
>> If someone can optimize the ctime code for speed and size, that changes
>> will
>> always be welcome; but I do not think that optimization for size should
>> be
>> done sacrificing/removing time zone features that other users expect to
>> be
>> provided by djgpp.
>
> BTW, on a semi-related note, a few years ago I mentioned that mkdir()
> was always pulling in ctime.o , so that will also make some apps even
> bigger in size. The culprit (for that separate case) is apparently
> xstat.c's _file_time_stamp(), which should probably be split into a
> separate file entirely (ftstamp.c ?).
>
> http://www.delorie.com/djgpp/mail-archives/browse.cgi?p=djgpp-workers/2009/09/26/21:32:54
>
> http://www.delorie.com/bin/cvsweb.cgi/djgpp/src/libc/posix/sys/stat/xstat.c
>

Another gem, yeah: mkdir() calling access() calling _is_executable()
needing _djstat_flags existing in xstat.c beside _file_time_stamp().
Ugghhhh..