Sneak some x86 test cases into the build and have it break. On Thu, 13 Oct 2022, 11:07 Tony McGee via ozdotnet, <ozdotnet@ozdotnet.com> wrote:
> My understanding is that you target what you want to support, so if he's > only shipping fully managed code and supporting x64 systems then it doesn't > matter too much but in that case may be sidelining x86 or ARM users for no > real reason. > It gets more complicated if you need to support multiple architectures or > have unmanaged dependencies where you need to match the bitness of the > dependencies with the application process, this is where AnyCPU and AnyCPU > (32-bit preferred) options will start to shine. > > https://dzone.com/articles/what-anycpu-really-means-net > > -Tony > > On 13/10/2022 08:45, Greg Keogh via ozdotnet wrote: > > Folks, one of my colleagues insists on compiling everything as platform > x64 mainly because he thinks "it's an x64 world and it creates a better > impression". For a year I've tried to convince him that for managed code > that it's a complete waste of time. I've told him that ildasm.exe shows > that for x64 and AnyCPU the generated IL and the manifests are identical, I > even told him that dumpbin.exe shows the only non trivial difference in the > PE headers is a couple of flags that show x86/x64 and PE32/PE32+, but they > don't affect the loading and running of a PE containing IL and metadata. > > Does anyone have paradigm-shattering evidence I can give my colleague to > break his habit? (I'm hoping I'm right of course!!) > > *Greg* > > > -- > ozdotnet mailing list > To manage your subscription, access archives: https://codify.mailman3.com/
