On Fri, Aug 25, 2017 at 08:39:29AM -0700, Tong Sun wrote: > I'm experiencing a *very very* strange problem now -- the same Go code is > producing different results *for me*. [...] > Can someone verify for me what you get please? > > go get github.com/go-dedup/fsimilar > > then > > cd go-dedup/fsimilar > go build > find test/sim -type f | ./fsimilar -i -d 12 -vv > > and tell me what's the last output that you got please. > > The problem is that *two *of my machines produce: > > [fsimilar] ## Similar items > map[Similars:[map[Hash:6184610222622303958 Dist:0 SizeRef:1 Name:GNU - > 2001 - Python Standard Library Ext:.pdf Size:1 Dir:test/sim/] map[Name:GNU > - Python Standard Library (2001) Ext:.rar Size:1 Dir:test/sim/ > Hash:6184610222622303958 Dist:0 SizeRef:1]]]. > *test/sim/GNU - 2001 - Python Standard Library.pdf* > *test/sim/GNU - Python Standard Library (2001).rar* > > But another one, the *only one*, produce: [...]
Two quick points: * What happens if you copy the "working" binary to the "wrong" machine and run it there? Does it work? I mean, you should rule out possible inconsistencies with build systems. This means both Go and all the graph of the libraries your project uses. If you're building on each machine, try doing clean-room building. This means freshly cloning the Go source code of, say, the 1.8 release branch and building it. Then copying over the whole hierarchy of the whole branch of the dependencies from the "working" machine into a new directory which must be a Go workspace. Then building there. * This bit [1] smells bad IMO. From the quick glance I failed to see where it's used but why it's there? What happens if you seed PRNG with a constant? 1. https://github.com/go-dedup/fsimilar/blob/master/fsimilarCLICmd.go#L70 -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
