Slashdot Mirror

OK. All of you have had a chance to rant, now is mine. He he he. :)

Although I am a relative newbie to slashdot, I have been reading it nearly every day since last October, and one thing has disappointed me. Unfortunately, most of the comments that I have seen, particularly on this article, happen to be not only anti-Microsoft, but also anti-user, in many respects. For the last 13 or 14 years, I have spent a good deal of time working with computers, mainly with an Apple IIe, an IBM PS/1, and my current machine, a PII-233 w/Windows 95. I know I will likely get flames from even mentioning that I even use Windows. Admittedly, there are many problems with Windows, and I don't like having to reboot at least every 24 hours. In fact, I am currently working on saving up for a nice box to throw Linux on and try it out. I am most anxious, but as a college student, I bought a Windows machine due to the fact that I was more familiar with it and the fact that the Systems Analysis department uses MS products for its classes on VB and C++. With that and books and other expenses, I am not in a position to buy the computer yet.

Yes, I use Office 97. I do like the integration, but there are several features that I would not miss if they are not there. Macros are one, Word HTML Authoring is another. I find it very frustrating that, after my having taken the proper precautions, that I read the comments here bashing practically anything even remotely related to Microsoft. I am a command line type of person, having been well versed in DOS, I have to admit that at times, I like the point and click simplicity of Windows, and I happen to enjoy the rich set of features that Outlook 97 (not Outlook Express) offers. The only lack is newsgroups, which I receive using Pine over a telnet connection to one af three different Un*x boxes around campus. I also like the PGP integration, although I could go without if I had to.

I also find it frustrating that some of the readers here assume that many who use Windows and Microsoft products are clueless and pathetic. Not all of us are. I would have to agree that many are, but not all. Microsoft products do have their merits, even if they are not perfect. I know that Linux isn't perfect. If anything was ever perfect, everyone would want it. That is simply not the case.

My question is this. "Linux is open source and, despite my use of Windows, I would have to agree that it is better, but why do we need to put down people that are not as literate as we are?" It would make sense that the Linux community would want to encourage as many people as possible to try it. For many people that post here, that is already their goal. For some of those and others that I have encountered in the Linux community around here, their means to the goal are misguided. The general impression that I sometimes get is that the Linux community is an elite crew that does not want more members, and some are inadvertently dissuading some users of Windows (even the die-hards) with the arrogant attitude that is projected.

Literate commentary welcome to come to nosbig@technologist.com. Please direct all flames to /dev/null.
Rob Gibson
nosbig@technologist.com
DH/DSS Key by email

I am needing a job; I just graduated with a CS/Math Degree last December. I wanted to see this article. I think, from the comments, what you say about headhunters is true. I have a few looking for me, and they keep finding tech support jobs. One guy even found a job doing backups for $10/hr. I thought my degree was supposed to be in demand (which is not the reason I got into this field, but it was a nice bonus).

Hey, I know this is a longshot, but if any of you know an entry-level programming or Internet/intranet development job in the Tulsa, Oklahoma area, let me know.

I have thought about this too, but there are a number of problems (there are certainly detailed in the papers about 4.4BSD union/null/portal fs).

First this is likely to break a number of programs that make assumptions about the FS semantics. You can have several times the same filename with differents inode. Removal of files if problematic: you remove a file, but then it can still be there in an inferior layer (I think that unionfs create a "blank" entry to mark that a particular filename has disappeared ; but then reboot and do the exact same mounts: bingo, the filename reappears).

Also sometimes the expected semantics can require heavy processing ; for instance if you just wish to have a CDROM mounted and a directory were modified files would be written, there would still be a problem: you may expect that the modified files would be transparently copied ; even if you don't require this semantics, you probably prefer that the non-existing parent directories in the path of your created file would be automatically created. But then what if they are symbolic links, etc... If you don't do that, then you have to manually mirror the whole directory tree of your cdrom: at that point you may realise that you may as well mirror the whole tree using symbolic links for non-directories, and that you won very little. If you change the cdrom, or if the partition, over which you did a mount, changes (say a NFS server serving /usr/local), you have to reflect the changes in your mirror-directories, etc... (looks like a nightmare)

So I think that even though union filesystems are very useful, they are only part of any solutions when they are used in real world (for packages versionning, pretending to write other a read-only partition [cdrom, workgroup NFS server], security by restricting the visible portion of the FS...). But maybe Inferno or {Open,Net}BSD or Hurd(?) users have more informations about using union filesystems in Real World.

An idea would be to experiment with a metalevel for filesystem namespace operations, i.e. for certain processes, each of their operations involving a filename could be forwarded to another process ("metalevel process" or something) that would decide how to handle this filename, and inform the kernel. I think that kernel modification (as opposed to glibc modification) is necessary, for operation atomicity (you don't want the filesystem to change because of another process, without being informed), and because only there inode manipulation can be done.

-- Cedric Adjih (adjih@technologist.com)