On Fri, August 17, 2007 1:11 pm, Andrew Lentvorski wrote: > > Given that the git Windows port is a second-class citizen, Git is not > bound to gather the most momentum. > > Mercurial specifically goes out of its way to support Windows. This > means that Mercurial, with time, is likely to achieve far greater > penetration than git. > > bzr also has a similar attitude as Mercurial toward Windows. >
This is a chronic problem which has 2 parts. Windoze lusers mindspace and real issues. Seeing as any modern system worth a tinker's damn is C/S, it really makes no sense to demand that the developers support more than one back-end OS. Maintaining more than one code line and supporting-server architecture in OSs that provide significantly different services is an invitation for one OS to become the buggy poor cousin, unjustly maligned (and, yes, it is unjust to malign windoze for not supporting fork). It's a back end, fer crissake. Fortunately most OSS SCM tools don't see the need to respond to this pressure. When they do, IMNSH opinion it's a mistake.[0] The problem, of course, is the mindspace of the windoze lusers and PHBs. "Oooo, what a shiny new SCM tool! It does everything we want ... but we _must_ have it for windoze servers. We're a windoze shop, you know." <simper> Front enda on windoze? Sure! All back ends? Gag me! The _real_ issue is platform dependent characteristics like EOLN and file-name case. Neither of these issues is insurmountable and both have been successfully addressed as far back as PVCS Version Manager, but oddly some current systems don't have the facilities to manage them gracefully. And IMNSHO this needs to be built into any system for it to be considered up-to-date. [0] BTW, for some tools like DevTrack, this Linux zealot openly supports using them on Windoze servers because (1) they're good tools, and (2) that's what they run on or run on best. -- Lan Barnes SCM Analyst Linux Guy Christian Seberino wrote: -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
