Working on the OS FAQ [Mega-Tokyo Wiki]

[BI lazy]

That's true; <verbatim> is a wiki-tag.

If you by any chance happen to take a glance at the html code produced by the wiki thing, you will be but so lucky to discover nice, juicy <pre> tags instead of the <verbatim> ones.

::)

[Solar]

...and <pre> as in HTML doesn't support Wiki highlighting. Both are rendered by PhpWiki, not by your browser.

The idea of Wiki is enabling people to write and edit who have no idea of HTML. Most of the time, that works very intuitive (like, space-indented text is rendered indented, a asterix-bullet list is rendered as <ul>, a word like *this* is rendered bold-fact etc.) - but if you want to label a whole text block to be rendered monotype, there's no simple way to do it, so they "fell back" to pseude-HTML tags.

Note that it is possible (and perhaps even default) to disable HTML tags in PhpWiki - the editing tips might just be wrong.

[Candy]

Thank you for reassuring my faith in humanity

I was a little nervous about again a new set of incompatible stuff.

one Q remains though, does Wiki generate HTML compliant stuff?

BTW, when/if I get the time I'll start helping out some more, it's just that I got this internship for the next 19 weeks (up to 25 june, my birthday), so before that I won't help... Will try to do anyway...

[Solar]

one Q remains though, does Wiki generate HTML compliant stuff?

You might want to test for youself - PhpWiki generates a "valid XHTML 1.0" button beneath each page. (Hint: Yes, it's compliant. )

[Solar]

This is a copy from the Wiki, since discussion is better done in the forum instead of the Wiki. Pype wrote in the "Microkernel" page:

In multiprocessors architecture, the drawback of context switching is also reduced, especially if the designer take care of keeping the 'common services' thread on a different CPU from its clients...

That, I am not so sure about. Sure, you can hold system services like networking and file system on different CPUs - but you would still have to switch around between them, whereas with a monolithic system you would not be residing on multiple CPUs, but execute the service request without switching involved... can you perhaps follow up that claim with some more info?

[Tim]

In fact, it is often worse in practice. If you've got more threads than CPUs, some threads will likely need to 'hop' between CPUs. This defeats CPU caching, because each CPU maintains its own cache. You can reduce CPU hopping by giving threads or processes affinity to a set of CPUs, but it's not possible to eliminate.

This is why a system with N CPUs doesn't run programs N times faster as a system with 1 CPU.

[bubach]

does this disscusion really belong in this thread??

[Solar]

Bear with us, we're just finding out how to best handle this. But you're right, I should have opened an individual thread for the OS-technical stuff.

[Candy]

In multiprocessors architecture, the drawback of context switching is also reduced, especially if the designer take care of keeping the 'common services' thread on a different CPU from its clients...

That, I am not so sure about. Sure, you can hold system services like networking and file system on different CPUs - but you would still have to switch around between them, whereas with a monolithic system you would not be residing on multiple CPUs, but execute the service request without switching involved... can you perhaps follow up that claim with some more info?

Well, if you got all kernel-certified threads not needing a full address space, in a monolithic kernel for 4+ CPU's, you could conceivably dump 'm all in one address space, and let that address space run on one CPU 100% of the time. Saves one hell of a lot of cache line invalidations etc

[bubach]

Ok, back to topic.. I added some FAT12 specific info to the FAQ..
Thanks to the person that cleaned it up..

/ Christoffer

[Solar]

I am currently working on a Wiki page that tells about the most often encountered licenses, and some of the issues involved when taking other people's code, verbatim or as tutoring material. This happens under "supervision" of Pype so I won't go on ranting (too much). ;D

Regarding "most often encountered licenses" - I considered writing about the following:

* GPL,
* LGPL,
* BSD,
* Public Domain,
* generic "commercial" license (EULA),
* code without any licensing statement.

My question is... anything I've forgotten? Meaning, of course there are dozens of other licenses in use, but does that list cover the basics?

[Pype.Clicker]

maybe a small 'what should i especially look at if the license is not one of these' ... in other words, things like 'i want to reuse files from a project under X in my project under Y, what should i look for in X and Y to know if i can do it ?'

[Pype.Clicker]

i was just walking the Pro-POS wiki and noticed a 'includepage' plugin ... so i couldn't resist in making a printer-friendly version of 'introduction' category ...

You're welcome to make such grouping for 'design', 'tools' and 'reference'

[Solar]

Pro-POS.org is going down in the foreseeable future. :-[

Virtually everything that was valuable information has been transferred to the OS FAQ some time before; but one thing has not yet been done.

At http://www.osdev.org/osfaq2/index.php/D ... %20Windows, I linked to a pre-made GRUB boot disk image located at http://www.pro-pos.org/downloads/grub_disk.zip. That file would have to be hosted somewhere else, or the link will go dead once the pro-pos.org domain is released.

Any volunteers? (Traffic isn't heavy on this one.)

[Pype.Clicker]

the .zip is now hosted on sourceforge, in 'miscellaneous' files for Clicker


https://sourceforge.net/project/showfil ... _id=240806

i guess i haven't any excuse left not to make Clicker booted by GRUB, now