On the 0x201 day of Apache Harmony Weldon Washburn wrote: > On 10/12/06, Geir Magnusson Jr. <[EMAIL PROTECTED]> wrote: > > > > Weldon Washburn wrote: > > > On 10/12/06, Geir Magnusson Jr. <[EMAIL PROTECTED]> wrote: > > >> > > >> > > >> > > >> Weldon Washburn wrote: > > >> > A word of caution to those who are committing C/C++ code. There are > > >> unique > > >> > features of Microsoft C/C++ that will cause a build failure on Linux > > >> and > > >> > vice versa. For example, gcc expects that the end of a C/C++ file is > > a > > >> > blank line. Microsoft does not. > > >> > > >> That's gotta be a bug. What version are you using? > > > > > > > > > gcc 3.4.5. If its a bug, its unlikely we will fix gcc or wait for gcc > > > to be > > > fixed. > > > > > Have you tried it on another version? > > No I have not tried it on another version. If Mike Ringrose is correct > > about the ISO spec, the basic problem is that MSVC compiler is not ISO > > compliant. I have hit this specific problem multiple times over the years. > > That is, code developed on windows laptop w/o a blank new line at the end > > will not compile on gcc. > > > No I have not tried it on another version. If Mike Ringrose is correct > about the ISO spec, the basic problem is that MSVC compiler is not ISO > compliant. I have hit this specific problem multiple times over the years. > That is, code developed on windows laptop w/o a blank new line at the end > will not compile on gcc. > > The real issue is that there are a bunch of "minor" incompatibilities > between MSVC and gcc. I would not recommend holding one's breath waiting > for them to go away. My approach to the above is to simply test on both > linux and windows before committing.
+1 for checking on both win&lin as a pre-commit (for C/C++). I recall many incompatibilities between the two compilers, so, checking both platforms is less of a burden than it is a benefit. Checking on various toolchains (gcc version, libc, binutils) seems not so obviously helpful to me. Because it is a heavy check, not giving much in terms of bug isolation. There are various real-life situations, when a toolchain version can influence on stability, but I think, they are not so common and are not worth the effort of checking before each commit. On compilation problems: let's take a more restrictive GCC (4.1?) and that would cook all compilation for us on Linux. -- Egor Pasko, Intel Managed Runtime Division --------------------------------------------------------------------- Terms of use : http://incubator.apache.org/harmony/mailing.html To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
