gnufx

53 ms·

631.
by gnufx 6y ago
"toxic wasteland" isn't the reason they gave for moving to Arm, though I don't have a reference to hand. I'm not sure about not missing a beat; as a component of Fukagu, this represents years of serious engineering
632.
by gnufx 6y ago
And specifically designed (along with the other system components) with suitable trades-off across the sort of work envisaged for it.
633.
by gnufx 6y ago
I don't think the original stored program computers even had files, but I don't know when the first recognisable filesystem was. There have been a number of image-based systems which you can program directly, possibly by direct
634.
by gnufx 6y ago
Darcs should be able to support semantic patches, but currently the only alternative patch type is token replacement. How useful semantics might be in a project with N different components' semantics, I don't know.
635.
by gnufx 6y ago
It didn't even require that. You fixed something from the immediately-buildable tarball that you'd rebuilt and tested, or even something that didn't need a rebuild (config, doc, interpreted source), and sent the change, perh
636.
by gnufx 6y ago
Hear, hear, except that Darcs' philosophy lives on in Darcs, thanks to Ben Franksen particularly, and I don't think the world needs anything else like Github. An "experimental" implementation of the Darcs 3 theory is in
637.
by gnufx 6y ago
This is probably not relevant to Jepsen, and I don't know how similar to a distributed database distributed metadata in parallel filesystems look. Anyway, some of the tests for the Lustre PFS use Linux' fault injection system. Als
638.
by gnufx 6y ago
Note it's by shivers, not pg (who I don't think was involved with T). It is worth reading the whole thing.
639.
by gnufx 6y ago
I feel that's not a good understanding of experimental physics. Why do you think I didn't understand the rather limited independent variables in my measurements or not understand the experimental setup well enough to debug experi
640.
by gnufx 6y ago
The important thing is to add error bars, which has probably been in XKCD passim. Yes, you want the numbers, but I've quite often just digitized plots from papers. I wrote a program for inverse plotting with an HP flatbed sometime in
641.
by gnufx 6y ago
I should have explicitly said the version 3 theory, which has an imlpementation in Darcs 2.16. I've never seen worst case quadratic complexity, like quicksort's, labelled exponential before, which seems misleading. I think there
642.
by gnufx 6y ago
Not to denigrate statistics at all, but this rings true to a physicist. In my field we won by increasing resolution and lowering background so the data were essentially obvious at some level, with small error bars from fitting peak areas.
643.
Xkcd: Statistics Tip(xkcd.com)
14 points by gnufx 6y ago|7 comments
644.
by gnufx 6y ago
How does he know? Was Dongarra completely misled, and how come it appears nothing like the Alphas I used, which certainly didn't just have a small scratchpad? Dongarra must have been pretty familiar with Alphas, of course.
645.
by gnufx 6y ago
To be fair, Roundy's idea of commutation was novel in 2002 (unless it was a reinvention of something earlier, which wouldn't surprise me) whether or not it was done right then. I'm happy if Pijul improves something which def
646.
by gnufx 6y ago
I'd still like to know if you're claiming Darcs 3 (which may or may not be at a similar stage of development) still has that issue (which I don't remember biting me). I've no particular axe to grind as a Darcs user, as
647.
by gnufx 6y ago
All this does seem to help make the argument for simple patch-based systems, i.e. Darcs and Pijul, in common with much discussion I see about git. (Their patches aren't just deltas, of course.)
648.
by gnufx 6y ago
Darcs is quite usable in my experience, for medium-sized projects, at least. Say ~50MB of source with ~10 years of history. Darcs has inspired the next version of itself (version 3), but it's not clear how that will compare with Piju
649.
by gnufx 6y ago
That's a significant bar, which doesn't apply to Darcs and Pijul.
650.
by gnufx 6y ago
Pijul patches can't be just diffs. I don't know the Pijul implementation but you can see what they're like in Darcs, for instance, in the text form they're meant to be mailed. (Darcs isn't alpha, nothing against P
651.
by gnufx 6y ago
I don't understand how this theory about Alpha and Sunway originated. Even if numerology suggested the Sunway chips were Alpha-derived, Dongarra[1] refutes it. TaihuLight is memory-constrained. The Alphas we used had enough cache to
652.
Scilab: Open-source software for numerical computation(scilab.org)
1 points by gnufx 6y ago|0 comments
653.
by gnufx 6y ago
Not to do down Octave, but if you want a free alternative to MATLAB, Scilab might be worth checking out too, particularly for an alternative to Simulink, as I understand it. It has a different emphasis, like not aiming at compatibility the
654.
by gnufx 6y ago
With Darcs you might put some tag in the patch name which would be caught by a hook if it was accidentally pushed.
655.
by gnufx 6y ago
I assume it's the same as with Darcs, where you would record a tag patch. The darcs --exact-version option prints that for a released version of darcs.
656.
by gnufx 6y ago
I'd have to check how it does/did it, but in the past I used the NHC (node health check) system for monitoring HPC cluster nodes. In this context the relevant check was for DIMMs dropping out. What it read needed a fudge factor,
657.
by gnufx 6y ago
Is pass-tomb not a decent solution to that? I've not considered it necessary in my threat model, so don't know.
658.
by gnufx 6y ago
Darcs has "replace" for that, the only other patch type that's been implemented for it as far as I know. I think Toolpack provided that for Fortran in the 1980s, but I don't remember for sure whether its version contr
659.
by gnufx 6y ago
For what it's worth, you can do that with Darcs import/export. I've kept parallel git and hg repos synced that way with a cron job; it could be done with a hook, but that slowed down pushes too much in my case. (Doing that
660.
by gnufx 6y ago
Pijul would presumably fit well with the development style encouraged by Sourcehut, if you need such a thing. I've never had time to consider integrating Darcs with it.

More ›