Disqus - Latest Comments for sgraham

Re: bash .inputrc tab completion incantations

sgraham — Fri, 04 Mar 2016 20:45:04 -0000

"\e[19~": history-search-backward

Re: bash .inputrc tab completion incantations

sgraham — Mon, 08 Sep 2014 23:38:25 -0000

and also

"\e[1;5D": backward-word
"\e[1;5C": forward-word
"\e[1;3A": "cd ..\n"

for ctrl-arrow and alt-up.

Re: Replacing the input reader in cmd.exe

sgraham — Sun, 11 Aug 2013 15:05:42 -0000

I'm sure many people do that. I find cygwin/bash janky, and the poor interop with other Windows programs is annoying.

Re: New Tab Page As Tasks

sgraham — Thu, 04 Jul 2013 11:47:49 -0000

Thanks Matt, I hadn't looked into V2 support yet. If you got it working, would you mind posting a link to your CWS version and I'll update the post to reference yours?

Re: New Tab Page As Tasks

sgraham — Wed, 13 Feb 2013 23:58:42 -0000

Instead of using this extension, you might be able to use Settings > On startup, and then put "https://mail.google.com/tas..." as one of the urls to open (or some variation on that).

Re: New Tab Page As Tasks

sgraham — Sun, 20 Jan 2013 21:33:30 -0000

Sorry, either Chrome or Gmail changed something for security reasons. Looking into it, but not sure if it's fixable or not yet.

Re: New Tab Page As Tasks

sgraham — Fri, 18 Jan 2013 18:05:14 -0000

There's been some changes to Chrome's NTP recently that broke it. Being investigated, but no workaround currently...

Re: Mini Jpeg Decoder

sgraham — Mon, 12 Sep 2011 19:36:01 -0000

Normally this is because you used gcc, not g++?

Re: New Tab Page As Tasks

sgraham — Wed, 13 Apr 2011 15:48:50 -0000

Sorry, no I can't. This is just a wrapper around the gmail
editor/viewer so I don't have control over sort.

Re: Kid's Games on the iPad - Miguel de Icaza

sgraham — Wed, 23 Mar 2011 19:31:24 -0000

Ah, my daughter is the same age. I didn't research much, assuming there'd be this sort of problem.

A recommendation list of "non-breakable" games would be appreciated, I'm sure.

Re: Mercurial (hg) with Dropbox

sgraham — Wed, 02 Feb 2011 12:55:28 -0000

The .hg directory is fully cloned when you use --noupdate. In Mercurial, .hg is the whole history of the repository of every revision. The working directory is just a copy of the current revision. When you clone on to another machine you get both the full .hg dir, and then the "update" gets a copy of the head revision to work on.

Re: Geohot releases PS3 jailbreak for firmware 3.55, world ceases to have any meaning

sgraham — Sun, 09 Jan 2011 01:12:45 -0000

Looking forward to XMBC now that we can get access to the GPU/SPU properly.

Re: Data Oriented C

sgraham — Fri, 07 Jan 2011 15:52:57 -0000

> "Books need to be written"...

Agree. (Of course you meant Cats are Felines are HairedMixin are Mammals are Animals are Objects. ;)

It's definitely not a perfect solution, in theory a different language would be a cleaner solution. I was going for a simple, small-increment improvement over what we have now.

I do imagine the body of the transform as basically similar to a shading kernel. I think it's somewhat different from shading languages/OpenCL, but really only because it's meant for the main body of code. It probably handles the packaging/invocation of the 'real' shading languages that might be used in the body of the transforms. (?)

It's a balance between allowing flexibility and constraining for the improvements that brings. e.g. We could make the transforms explicitly 1:1 (map) and you only have access to one element of the collection at a time in the transforming kernel. Or only specifying a reduction transformation. Then they could be auto-parallelized/distributed, but this of course creates the problem of what to do when you need access to other elements in the collection.

Maybe that's how to make the default right, map/reduce/whatever is the default, and a generic "transform" is more advanced fringe functionality.

I *do* think it's important to stay close to C, even though it'd be nice not to. There's just no chance of actually achieving adoption without a Microsoft or Google-sized investment otherwise.

Re: Data Oriented C

sgraham — Fri, 07 Jan 2011 13:40:10 -0000

I think unit tests are mostly orthogonal, but maybe it'd make sense to have a `test` block in there.

Agreed, `input` should be immutable.

Not sure what the platform-specific optimizations look like, but definitely agree. I think there would probably end up being extra markup or keywords to specify things like alignments, padding, or whatever. Some of it might be defaulted properly based on sku, but at this level you're always going to want to override some of it. I'd want the last chain of the pipeline to be able to output to other subsystems for interop (anim, ik, render buffer), so you want precise control.

Re: Windows Phone 7 'NoDo' update hitting phones in early February, 'Mango' coming later with IE9?

sgraham — Tue, 04 Jan 2011 19:25:34 -0000

Wow, that's a bad code name. "No Do". Whatever Droid Does, etc.. Reminds me of that story about the Chevy Nova in Spanish speaking countries.

Re: AirSync for doubleTwist brings wireless syncing to Android phones

sgraham — Tue, 30 Nov 2010 20:43:54 -0000

Is this somehow better than Subsonic? Works pretty well for me.

Re: Disable STYLE-WARNING on sbcl

sgraham — Mon, 29 Nov 2010 00:27:45 -0000

Sorry, not sure, I've never used CLISP.

Re: Sarah Palin’s Kids: The Complete Lack of Online Self-Control Doesn’t Fall Far from the Tree

sgraham — Thu, 18 Nov 2010 22:34:55 -0000

whose appropriate?

Re: New Tab Page As Tasks

sgraham — Wed, 13 Oct 2010 23:09:29 -0000

Not that I'm aware of, opening the "main" page seems to be hardcoded into Tasks.

Re: Skulpt Parsing Revisited

sgraham — Fri, 20 Aug 2010 12:34:41 -0000

Hi Mark, I agree. Next post will be how the compile phase was rewritten, and 'import' is finally well supported.

My next goal is to make goog.* trivially importable (goog.ui.* and goog.graphics.* is what I want, personally). I haven't started working on that yet, but my hope is that it won't be too complicated as Closure Library is quite static and doesn't play sneaky JavaScript tricks.

Re: Opcode Design for a Lisp processor

sgraham — Tue, 15 Jun 2010 21:39:48 -0000

Yeah, I'm having fun playing with this stuff. :)

I've been avoiding Verilog and VHDL, they seem a bit mucky, but I'll have to hunt around for some good tools and tutorials. Probably more realistic and effective than writing my own simulators anyway, if I put some time into learning how to use them.

Re: Not Invented Here by Bill Barnes and Paul Southworth

sgraham — Tue, 15 Jun 2010 03:26:14 -0000

Ah, I get it now. :)

Re: New Tab Page As Tasks

sgraham — Sat, 03 Apr 2010 21:55:10 -0000

Yup, I can do Apps support. Does anyone know what the url for tasks in gmail
for apps is? (Not a user myself...)

Re: New Tab Page As Tasks

sgraham — Thu, 11 Feb 2010 16:58:58 -0000

Sorry, I don't really use FF any more, so not by me.

Re: Configuring Vim, some more

sgraham — Tue, 11 Nov 2008 12:47:57 -0000

re: noerrorbells

Oh, you're right. I tend to disable the 'default beep' completely in the OS so I didn't notice.