smart_str_appendl() would be faster since it'll avoid having to
calculate the string length, and sizeof() can be used to determine the
length of the constant, resulting in faster code.

On a general note, I am not sure its a good idea to change the EOL for
CSV file, since many scanners expect "\n" as a line separator and the
change may introduce regressions.

Hi Ilia,

Sorry for my C confusion, like I said it's been awhile.

Anyways, would smart_str_appends() be the correct function to use
then?

John Mertic
jmertic@gmail.com
http://jmertic.wordpress.com

"Explaining a joke is like dissecting a frog: you understand it
better, but the frog dies
in the process." --Mark Twain

On Wed, Oct 22, 2008 at 4:16 PM, Ilia Alshanetsky ilia@prohost.org
wrote:

You cannot use smart_str_appendc() in this case, since the EOL
could be a 2
byte string "\r\n".

smart_str_appendl(&csvline, PHP_EOL, sizeof(PHP_EOL)-1); should be
used
here.

Hi Lars,

Thanks for the pointers, updated the patch and added a test.

Index: file.c

RCS file: /repository/php-src/ext/standard/file.c,v
retrieving revision 1.530
diff -u -r1.530 file.c
--- file.c 21 Oct 2008 22:06:48 -0000 1.530
+++ file.c 22 Oct 2008 18:21:10 -0000
@@ -2104,7 +2104,7 @@
}
}

•   smart_str_appendc(&csvline, '\n');
  

•   smart_str_appendc(&csvline, PHP_EOL);
  smart_str_0(&csvline);
  
  ret = php_stream_write(stream, csvline.c, csvline.len);
  

John Mertic
jmertic@gmail.com
http://jmertic.wordpress.com

"Explaining a joke is like dissecting a frog: you understand it
better, but the frog dies
in the process." --Mark Twain

On Wed, Oct 22, 2008 at 10:36 AM, Lars Strojny lstrojny@php.net
wrote:

Hi John,

Am Mittwoch, den 22.10.2008, 10:03 -0700 schrieb John Mertic:
[...]

Below is a patch to fix this issue; it uses the constant PHP_EOL
to
get the correct newline to use on the current platform:

Thanks for your patch. A few things to mention, as it is your first
patch: please use "diff -ru" to create unified diffs. Also we try
to
always add tests for the things we fix or create. Would you mind
creating a test for the fix you sent to make sure no regression
happens
in the next n years?

Thanks, Lars

Jabber: lars@strojny.net
Weblog: http://usrportage.de

--

Ilia Alshanetsky

Ilia Alshanetsky

You cannot use smart_str_appendc() in this case, since the EOL could be a 2
byte string "\r\n".

smart_str_appendl(&csvline, PHP_EOL, sizeof(PHP_EOL)-1); should be used
here.

Hi Lars,

Thanks for the pointers, updated the patch and added a test.

Index: file.c

RCS file: /repository/php-src/ext/standard/file.c,v
retrieving revision 1.530
diff -u -r1.530 file.c
--- file.c 21 Oct 2008 22:06:48 -0000 1.530
+++ file.c 22 Oct 2008 18:21:10 -0000
@@ -2104,7 +2104,7 @@
}
}

•   smart_str_appendc(&csvline, '\n');
  

•   smart_str_appendc(&csvline, PHP_EOL);
   smart_str_0(&csvline);
  
   ret = php_stream_write(stream, csvline.c, csvline.len);
  

John Mertic
jmertic@gmail.com
http://jmertic.wordpress.com

"Explaining a joke is like dissecting a frog: you understand it
better, but the frog dies
in the process." --Mark Twain

Hi John,

Am Mittwoch, den 22.10.2008, 10:03 -0700 schrieb John Mertic:
[...]

Below is a patch to fix this issue; it uses the constant PHP_EOL to
get the correct newline to use on the current platform:

Thanks for your patch. A few things to mention, as it is your first
patch: please use "diff -ru" to create unified diffs. Also we try to
always add tests for the things we fix or create. Would you mind
creating a test for the fix you sent to make sure no regression happens
in the next n years?

Thanks, Lars

Jabber: lars@strojny.net
Weblog: http://usrportage.de

--

Ilia Alshanetsky

Hi Ilia,

I can see your point about scanner's having trouble with different EOL
characters, but I know in our use case the problem is that the CSV
files created will come out the lines all mashed together, which is a
problem for readability of the file. And since most CSV scanners know
how to deal with Windows, I would think we are pretty safe ;-)

Attached is an updated patch that uses this function instead.

--
John Mertic
jmertic@gmail.com | http://jmertic.wordpress.com

Hi All,

I haven't seen any further comments on this bug; does it seem safe to
commit or should it wait? I've re-attached the patch and test again
for everyone's reference.

Thanks!

John Mertic
jmertic@gmail.com | http://jmertic.wordpress.com