[Mono-devel-list] mod_mono on osx / big endian systems.

grompf grompf at sublimeintervention.com
Wed Jun 9 16:54:31 EDT 2004


   On second thought for "portability", it would probably be better to 
determine endianess in mod_mono.c, and do the conversion there as 


On 9-Jun-04, at 4:42 PM, Bernie Solomon wrote:

> The "big endian" in the comment made me look at this not that I
> use mod_mono but do use big endian machines.
> Shouldn't this be a run-time check not a compile time one
> so you don't need to rebuild DLLs. BitConverter.IsLittleEndian
> can be used to test the endian-ness of the platform.
> Bernie Solomon
> ----- Original Message -----
> From: "grompf" <grompf at sublimeintervention.com>
> To: "Mono Development" <mono-devel-list at lists.ximian.com>
> Sent: Wednesday, June 09, 2004 1:29 PM
> Subject: [Mono-devel-list] mod_mono on osx / big endian systems.
>> I've been debugging the usablilty of mod_mono on OSX with Apples 
>> Apache
>> 1.3.  Out of the box this doesn't work; when making a request
>> mod_mono_server uses 100% cpu and locks.  I tracked the problem down 
>> to
>> endianess:
>> BinaryReader.ReadInt32 assumes little endian on the stream; which for
>> most of .Net is fine; however since mod_mono is native, and writing 
>> the
>> in to the stream; this is wrong.  Meaning that ReadString() in
>> ModMonoRequest.cs is getting some insane size for the string to read.
>> This patch solved the problem locally; but probably isn't ideal.
>> This patch implements its own ReadInt32 which is different depending 
>> on
>> wether BIGENDIAN is set or not...
>> -kangaroo
> !DSPAM:40c775e493079028320949!

More information about the Mono-devel-list mailing list