I'm having some troubles getting gettext work on a Linux server. It works on my Windows machine. So one of the possible reasons I found is that MO files might be platform specific.
In this article, it's written:
Since the .mo files are binary, they are platform specific.
But is that really the case? Why would the gettext team make them platform specific? That would be like saying "Since .jpg files are binary, they are platform specific", which thankfully is not true.
So does anybody know anything about this? I can't find much information on gettext's reference documentation.
Looking at the gnu gettext source it at least accounts for different endiannesses and uses fixed 32 bit ints. The file format is defined in the file gmo.h. So I think gnu gettext files should be portable. However, there are also other gettext implementations. Since the article you linked mentioned solaris it might be that the story is different there.