Re: [upnp-sdk-dev] Maintenance and large file patch

[Peter E. <sf...@en...>]

As a caveat to my other comments; I have a fair background in=20
programming but am an absolute n00b on Linux so I've been on a rather=20
steep learning curve so far!

R=E9mi wrote:
> On Saturday 22 April 2006 07:44, Peter Enzerink wrote:
>  =20
>> I have just posted a patch for 64 bit file support. It includes a
>> change to upnp.h which affects programmes using the library. This is
>> necessary because both the file length and offset needed to be
>> increased in File_Info.
>>    =20
>
> thanks, this is good news !
> The patch looks good, I have some comments below.
>
> Also : your patch corrects only the server half (webserver parts).=20
> Needs to be done as well : the client download functions (Upnp*HttpGet*=
) : are=20
> you willing to address this as well ?
>  =20
I hadn't looked at the client but will do so.
>> Well almost! The filesize is handled properly by the browser interface
>> but CDS is suddenly bombing.
>>    =20
>
> which CDS ?
>
> Best regards,=20
> 	R=E9mi
>  =20
It was just one of a number of areas where the return type was too=20
small. The soap code had more of them. I believe they have all been=20
fixed in the server code.
> -----------------------------------------------------------------------=
------
>
> - general : avoid including generated files (Makefile.in) in the patch,=
 it=20
> becomes difficult to read
>  =20
Ah yep.
> - don't forget to patch the ChangeLog as well
>  =20
I didn't want to make this an official change until people more=20
knowledgeable than me could provide feedback.
>> diff -Naur libupnp/upnp/Makefile.am libupnp-1.3.1/upnp/Makefile.am
>> --- libupnp/upnp/Makefile.am	2006-03-06 07:31:17.000000000 +1100
>> +++ libupnp-1.3.1/upnp/Makefile.am	2006-03-20 11:39:09.000000000 +1100
>> @@ -9,7 +9,7 @@
>>
>>  AM_CPPFLAGS		=3D -I$(srcdir)/inc \
>>  			  -I$(top_srcdir)/threadutil/inc \
>> -			  -I$(top_srcdir)/ixml/inc
>> +			  -I$(top_srcdir)/ixml/inc -D__USE_FILE_OFFSET64
>>
>>    =20
>  =20
> hmm. This shouldn't be necessary : the AC_SYS_LARGEFILE check I already=
 put in=20
> configure.ac normally sets in a portable way all the needed flags for=20
> largefile offsets (see generated autoconfig.h :  _FILE_OFFSET_BITS,=20
> _LARGEFILE_SOURCE ...). In particular with glibc, __USE_FILE_OFFSET64 i=
s=20
> automatically set if _FILE_OFFSET_BITS is set to 64.
> Did you have problems without this ?
>  =20
I could be wrong but I think that diff was in the CVS version of the=20
code. If it isn't then it probably isn't required anyway.

Cheers