IWETHEY v. 0.3.0 | TODO
1,095 registered users | 0 active users | 0 LpH | Statistics
Login | Create New User
IWETHEY Banner

Welcome to IWETHEY!

New Cache preload idea
I started thing about ram/vs/disk performance issues and
occasional pauses being the cost of the benefits of
swap space.

Here: [link|http://kerneltrap.org/node/view/3202|http://kerneltrap.org/node/view/3202]

The shell often "knows" what you are about to work on.
You sometimes tell it by command line completion. But
it could do more.

If the system load was low enough, it could preload your
programs and data into memory (as you are typing them) to
give the appearance of almost a pure ram based system for
most command line driven interaction.

As you type a command, your shell could examine your
history of commands and trigger the kernel to load into
the page cache the most likely program(s) that you are
about to execute.

Maybe it won't bother until it hits the space for programs
that take command line argument, but if your historical pattern
for 'gvim' usually did not include an argument, then it could
have loaded it before you hit return.

It does not need to search your path for new commands, only
those in your command line history. But it should keep a
rather large list since the goal here is to reload something
that probably is not in memory due to the fact it has been
a while since it was executed.

As you hit a space, it should move into data mode, trying to
predict what file(s) you are about to access and preload them
as well. Pipe? Back to program checking.

Load wrong? Need a method to notify the kernel that it really
doesn't care about certain loaded prgram / files.

Is this a stupid idea or worth fleshing out?
New OT: Stop pressing return at the end of every line, dammit!
:-)


Peter
[link|http://www.debian.org|Shill For Hire]
[link|http://www.kuro5hin.org|There is no K5 Cabal]
[link|http://guildenstern.dyndns.org|Blog]
New No
Be happy I don't write in ALL CAPS!

If this annoys you so much that the value of my posts are not
worth it, then killfile me. I don't consider this abusing you or
your screen.

I'm sorry you feel it displays poorly. Either it will leave a bunch of
space to the right, which is not too awful, or it will leave a bunch
of jagged wrapped lines when you wrap before the spot I intended,
and that would suck.

I'm comfortable at about 50-60 characters per line for conversational
text. Hitting return at about that point ensures it.

But if that causes jaggedness, tell me what your typical line width
is and I'll try to adhere to it.

Sometimes, depending on the web site, depending on the variety of other
stuff shown, the automatic wrapping does not work. When the text is
displayed, it ends up constructing a box just a LITTLE bit wider than
the browser. If I make the browser wider, the box grows to match.
No matter how wide I make the broswer (and I have a 2 screen display),
I always have to use the LEFT<->RIGHT scrollbar on the bottom of the
screen.

This happens in both Mozilla and Konquerer to me.

One (but just one) of the reasons I hit Enter is to avoid that. Others
include the fact I often write posts in gvim and then cut&paste them in.
I always have hard returns in there. Yes, I am aware I can fiddle with
the pasting to join together continuous lines and drop the hard returns.
I don't want to!
New It's hard to read. Thought I'd ask. Shrug.


Peter
[link|http://www.debian.org|Shill For Hire]
[link|http://www.kuro5hin.org|There is no K5 Cabal]
[link|http://guildenstern.dyndns.org|Blog]
New I'm not sure that it is hard to read
There is a reason that newspapers have columns of print of the width that they do. That is a convenient width for people's eyes to scan. Barry's style is inconvenient to scroll past, but to read?

Cheers,
Ben
To deny the indirect purchaser, who in this case is the ultimate purchaser, the right to seek relief from unlawful conduct, would essentially remove the word consumer from the Consumer Protection Act
- [link|http://www.techworld.com/opsys/news/index.cfm?NewsID=1246&Page=1&pagePos=20|Nebraska Supreme Court]
New Re: I'm not sure that it is hard to read
I think it must be because other comments are not narrower. On reflection, I think that the visual break between "full width" comments and, say, Barry's comments produces a break in my initial expectation of where my eye thinks the text should flow to.


Peter
[link|http://www.debian.org|Shill For Hire]
[link|http://www.kuro5hin.org|There is no K5 Cabal]
[link|http://guildenstern.dyndns.org|Blog]
New I'd be interested
Reading that thread I also heard about swappiness. Dang, but that sounds like a useful parameter that I wish I had heard about earlier. I hate coming to a machine in the morning and finding that all of my programs have been swapped out for the convenience of a cron job which was not performance critical...

Cheers,
Ben
To deny the indirect purchaser, who in this case is the ultimate purchaser, the right to seek relief from unlawful conduct, would essentially remove the word consumer from the Consumer Protection Act
- [link|http://www.techworld.com/opsys/news/index.cfm?NewsID=1246&Page=1&pagePos=20|Nebraska Supreme Court]
New That reminds me of "time sharing" systems of old.
All the users would be on dumb (and mechanical) terminals like an [link|http://www.columbia.edu/acis/history/1052.html|IBM 1052]. People would fiddle with the shift key while reading the output on paper. The activity would keep the user in a higher priority queue.
Alex

Honor has not to be won; it must only not be lost. -- Arthur Schopenhauer (1788-1860), German philosopher
New That page links to some very cool stuff
Primarily, [link|http://www.nfrpartners.com/comphistory/|photos of old IBM equipment]. I remember as a work experience kid having to lean the vacuum channels of the tape drive - gangs of fun.

The author's wry commentary is good value too.

Picture A : "Decks of punched cards"
Picture B: "Insert the cards at left to produce these. They are called 'core dumps'" :)

And the wall behind the 1130 system is just ... far out.
Two out of three people wonder where the other one is.
New Everyone smoked.
New Thanks, I hadn't seen those.
That was my world in my late 20's, early 30's.
Alex

Honor has not to be won; it must only not be lost. -- Arthur Schopenhauer (1788-1860), German philosopher
New Bits of it were my late teens, early 20s.
I was just a mainframe user, not an operator. (My week of work experience as a 15yr old notwithstanding.)

Of course, being a young mainframe COBOL programmer generally meant I hated it, longing for something more current. But nowadays I look back on it with a certain fondness. There are still aspects of ISPF that I miss. Well, a bit. Maybe.

(I don't miss having to carry armfuls of tape reels across site to the 'backup safe'...)
Two out of three people wonder where the other one is.
New Interested? Then comment
Tell me what is good, bad, or stupid!
Point me into area I will need to research before thinking about devleoping.
Give me the email of someone who might actually do it, rather than me!
New Well...
I like preloading programs. Those tend to be reasonably often accessed, and of reasonable size. Though some might dislike it because they say, "I load a program, and part of it never gets modified, why not just keep only what I need paged into memory?" My gut says that this is wrong, but some good benchmarks would be useful.

I don't like guessing at files though - constantly guessing wrong would massively increase the size of your working set, causing you to make the swap problem worse for other programs.

You might achieve a modicum of sanity by setting limits on how much data/program you are willing to preload. So you increase the working set, but by a tunable amount.

Cheers,
Ben

PS About an email address for someone that I can think of who might do it, do you really want me to give you your own email address? :-P (My way of saying that nobody comes to mind.)
To deny the indirect purchaser, who in this case is the ultimate purchaser, the right to seek relief from unlawful conduct, would essentially remove the word consumer from the Consumer Protection Act
- [link|http://www.techworld.com/opsys/news/index.cfm?NewsID=1246&Page=1&pagePos=20|Nebraska Supreme Court]
Expand Edited by ben_tilly May 31, 2004, 02:06:08 AM EDT
New The horror of MS Office FastFind springs to mind
...there was a hideous model of anti-efficiency if I ever saw one.
New What do you meam: WAS?
You mean STILL is. It lives on in the OS called "Volume Information Index" (or something close)

No frickin way does ANYTHING need to know as much about the filesystem. Or the contents of the files. Sheesh.
--
[link|mailto:greg@gregfolkert.net|greg],
[link|http://www.iwethey.org/ed_curry|REMEMBER ED CURRY!] @ iwethey

Give a man a match, he'll be warm for a minute.
Set him on fire, he'll be warm for the rest of his life!
     Cache preload idea - (broomberg) - (15)
         OT: Stop pressing return at the end of every line, dammit! - (pwhysall) - (4)
             No - (broomberg) - (3)
                 It's hard to read. Thought I'd ask. Shrug. -NT - (pwhysall) - (2)
                     I'm not sure that it is hard to read - (ben_tilly) - (1)
                         Re: I'm not sure that it is hard to read - (pwhysall)
         I'd be interested - (ben_tilly) - (9)
             That reminds me of "time sharing" systems of old. - (a6l6e6x) - (4)
                 That page links to some very cool stuff - (Meerkat) - (3)
                     Everyone smoked. -NT - (broomberg)
                     Thanks, I hadn't seen those. - (a6l6e6x) - (1)
                         Bits of it were my late teens, early 20s. - (Meerkat)
             Interested? Then comment - (broomberg) - (3)
                 Well... - (ben_tilly) - (2)
                     The horror of MS Office FastFind springs to mind - (FuManChu) - (1)
                         What do you meam: WAS? - (folkert)

Come sweet slumber, enshroud me in thy purple cloak.
65 ms