On 4/24/12 13:13, "Pete Batard" <p...@akeo.ie> wrote: >On 2012.04.24 07:38, Kustaa Nyholm wrote: >> Thanks,but I think I was just curious for the actual technical issues >> those new fields will cause. > >Well, to me the technical issue, is that in its current instance, >Peter's patch completely ignores MS build environments, as well as any >environment where git is not invokable by default. For instance, someone >may fetch the repo on one platform and compile it onto another where git >is not present - not a far fetched scenario for a secure corporate >environment where internet access is limited to a few machines, or where >libusb would be used with an internal distributed VCS that is different >from git.
Ah, so Peter's 'solution' (as I've said I've been tool lazy to look at the code) to some perceived problem requires that the build environment invokes git ?! Looks pretty broken to me. > >In my view, this "breaks" cross platform because, unlike our current >versioning, if we use this field alone, we end up with platforms where >we won't get the full version data, and --I hope this'll help see where >I'm coming from-- where the exact same codebase produces a different >output for something that should and (as demonstrated) can be the same >on all platforms. Agree. > >Now, as Hans and others suggested, one may also consider that the only >technical issue that matters besides compilation errors, if we don't >carry the field, is the crash apps compiled against the libusb shared >library are likely to experience when trying to access this data. And >this can indeed easily be prevented by adding an extra static string to >the version. > >Right now I'd say the majority seems to see the latter approach as both >low risk and preferable, so we're likely to go with that. Given >Michael's position however, I can't entirely eliminate some concerns >about this extra field being somehow used as an attempt to ditch what I >see as a better, as well as actually cross-platform, approach to >versioning. Ok, I can live with those fields filled with some constant values, if that prevents crashing and the application otherwise runs fine. No *strong* feelings either way, but they do seem kind of stupid and solve no real problem because code that uses them is non existent atm and hopefully and likely not many people will use them ever, and thus it just something stupid we will carry on forever. br Kuti ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ libusbx-devel mailing list libusbx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libusbx-devel
