On 06/10/12 13:16, Jakub Lach wrote: >>> Sometimes it would be enough just to >>> test if the port compiles before committing it (I'm talking about >>> libreoffice here which is broken) > > Yeah right... Like updating libreoffice without testing > would be actually possible at all... > > Are you familiar with http://redports.org/ ? > > Just a example what is used for WIP. > > -- > View this message in context: http://freebsd.1045724.n5.nabble.com/Re-Why-Are-You-NOT-Using-FreeBSD-tp5714183p5717150.html > Sent from the freebsd-current mailing list archive at Nabble.com. Well, this thread has been run out of the subject and is now being hijacked by port's problems ;-) One point to add for the NOT using is definitely the mess in the ports (sometimes). I have the feeling that, for my now almost 16 years experience and usage of FreeBSD, many statements PRO FreeBSD and the ports are repeated from times, when FreeBSD's port system was superior over many other approaches. Time changes, others keep pace. Just to comment on that. FreeBSD is supposed to have its strength in server environment. But I see lack of those strength when it comes to user management, specifically OpenLDAP. For users it is not very convenient changing their password via a cryptic "ldappasswd" statement. On most Linux boxes I have access to, (Suse, Ubuntu), one can simply use the system's "passwd" command - everything else is then done via PAM. I had in 2007 and 2008 some issues with that and simply by that fact, FreeBSD was banned from the desktop side and after it has been banned by the company I worked for from their desktops, they didn't see reasons why they should put more effords into administering server based on FreeBSD AND desktops based on Linux. The decission then was made to use Linux Ubuntu server and desktop. One more pebble erodet to dust ... That was an example of a data mining business company. Years ago the physics department at my former university had their networking infrastructure based on FreeBSD - that was in the time of FreeBSD 4. Strong network stack, stable, easy to manage. With Linux kernel 2.6 most of the PROs for FreeBSD where obsoleted, and as shown here, FreeBSD does have some disadavntages in throughput compared to Linux. After those benchmarks have been publsihed things may changed, but the "negative" information against FreeBSD is sticky. Or, in the opposite way, remnant informations of the past "PRO" FreeBSD are sticky (a good luck then for FreeBSD). Talking about NOT using FreeBSD, for me there is a bunch of reasons to change and these reasons have a gravity to my profession and work. FreeBSD unluckily lacks in optimized mathematical libraries and compilers. While I'm happy to live with LLVM/CLANG and GCC 4.6 or GCC 4.7 and it's Fortran derivatives (I don't use Fortran, but need to compile some model software written in F95), others don't. A complete No-Go is the lack of CUDA and more important OpenCL capabilities and therefore GPGPU usage. As nVidia made clear in San Jose, CUDA, and therefore GPGPU, is a tremendous fast growing market. On all of our number crunchers we use now Linux - for exactly this GPGPU reason. And once seddled, I guess it is hard to convince people to move towards another OS. I have no clue how to change this. It is a political issue beyond my capabilities. I'd like to see more advertising FreeBSD or any *BSD in the scientific development, but it seems everything is stuck to Linux - because it is the better and faster OS (also something that is often brought up without evidence, but it is stick in the heads). Who ever has ordered hardware from Dell and tried to manage their JAVA based blade accessing modules via native FreeBSD applications, knows that it is a pain in the ass. For just checking the HPC servers from time to time I need to use Windows or Linux (most run in a VBox with a crappy screen). If the advantages of your favorite OS does not give you a tremendous massage of your feelings, I guess those issues will make you turn towards what is more convenient much faster.
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:27 UTC