Last time i tried openSSL didn't compile, zlib did afther some tinkering with it.
Any performance stats on the Win64 build? mine where always worse than the Win32 builds (using ab.exe)
On 4/26/06, William A. Rowe, Jr. <[EMAIL PROTECTED]> wrote:
Mladen Turk wrote:
> Jorge Schrauwen wrote:
>
>> wait... did you do a 32bit binary on a win64 system or a 64bit binary?
>>
>
> Microsoft Windows Server 2003 R2
> Standard x64 Edition
> Service Pack 1
> AMD Athlon 3600 2.20 GHz, 2 GB of RAM
>
> Miscrosoft Visual Studio 2005
> Version 8.0.50727.42
>
> Lots of warnings mostly size_t to int conversion,
> but it works and compiles.
>
> Be careful when building OpenSSL.
> I'm using:
>
> perl Configure VC-WIN64A
> call ms\do_win64a
> nmake -f ms\ntdll.mak
> mt -nologo -manifest out32dll\libeay32.dll.manifest
> -outputresource:out32dll\libeay32.dll;2
> mt -nologo -manifest out32dll\ssleay32.dll.manifest
> -outputresource:out32dll\ssleay32.dll;2
> mt -nologo -manifest out32dll\openssl.exe.manifest
> -outputresource:out32dll\openssl.exe;1
This raises a *great* question. Do we wish to include manifests or drop
this with /manifest:no for our builds? I'm still trying to grok an advantage
to this 'feature'. The benefits are obvious for .NET applications but i've yet
to find an advantage to using them for native code.
Bill
--
~Jorge
