1. You can pick a mirror relatively trivially, but since I've run the program, the fastest one isn't the one I chose manually. Also, it can choose multiple mirrors at once, so presumably if there is a failure, it will choose the next mirror(s) that it wrote down in pkg.conf
2. You are saying that the ftp protocol can be implemented trivially? You are ridiculous sir. 3. How do you suggest I filter out obviously bad choices. Add on a perl geolocation package that isn't available in a base install. How about I just ftp download a smaller file to discover the latency. 4. How doesn't it meet standards. I wrote it according to the style man page as far as I can tell. And I ran it through indent. Even though I think kernel normal form is less readable. I think that there is some unwritten policy that nobody can get something like this into the system. Why on earth hasn't this happened yet? On Feb 1, 2016 10:48, "Dmitrij D. Czarkoff" <czark...@gmail.com> wrote: > Jorge Castillo said: > > Why not make it a port? > > Making port for figuring out PKGPATH doesn't sound right. > > See, there are four problems with the program: > > 1. It is not good enough in doing its job. Which is funny, because > picking right mirror is trivially done without any program. > 2. It uses external tools for tasks that could be trivially implemented > in C. > 3. It doesn't filter out obviously bad choices, eg. users in Europe > will test mirrors in North America. > 4. It doesn't meet OpenBSD's standards for code in base. > > Problems #2, #3 and #4 can be fixed, but problem #1 makes this > discussion completely pointless. Provided that all OpenBSD developers > who cared to participate in this discussion pointed out this issue, I'd > suggest to stop wasting time and bandwidth right here. > > Luke, if you disagree with my assessment, please publish your program on > github and convince tech media to mention it. And move to next thing. > Thank you in advance. > > -- > Dmitrij D. Czarkoff >
