goofy are you still working in game dev?

We had a required class in C# @ university and I was actually positively surprised. The language isn't bad and .net is half decent as well

Yeah.

.NET is great until you run into something that is broken in the framework, or where the framework has goofy limitations because the .NET component is just a wrapper around some base win32 bs, and then it can be hell to work around

I have a bug right now where in some ******ed corner case the operating system doesn't send WM_PAINT events to a control in a tab.

It's driving me nuts

Tell me your troubles well named. I'm old friends with WM_PAINT and all of the system controls (user32 ones -- comctl is more acquaintances)

Also, run spy to see what happens?

lol +1 to old job and spss.

ported some old vax/vms fortran code to linux machines when i was a freshman in college

we use git at work, but I'm still learning. I liked cvs

ffffffffyyyyyyyyyeeeeeeaaaaaaaahhhhhh


No, but really I'm a C or C++ coder (ok, not really a coder. a mathematician who can code) with some experience in Java and about 20 other languages/environments (among others: mathematica, matlab, maple, magma, sage, ada, fortran, smalltalk, lisp, scheme, pascal, html, and some bash scripting ldo) that just took a job where everything's written in a homegrown python/cython variant. So now I'm learning that.

browser question for you smart guys

on my phone, I often miss the arrow for go to new posts and click the main link which takes me to post 1 of the thread but if I click the arrow fast enough if will follow the 2nd command, what decides the cut off point of which command to follow?

probably it doesn't tear down the document for previous page until a certain point in the load process for the page it's downloading

If you click again before it disposes the document the click handler works and overrides the request with the new one. If you click too late you're not actually clicking on a "link" anymore

It appears that over the summer I became the API guy at work

Every time we need to integrate some outside service, I'm the one that they toss the vague requirements and a link to documentation to. APIs I've worked on since June:

- Google Data/Docs (to create/edit/import/etc Google docs)
- Google's OpenID implementation
- Google Calendar
- A couple different PDF ones until we finally settled on one that worked
- Amazon web services (**** YOU)
- Vuzit, which is a pretty sweet online document editor, similar to G Docs
- zencoder, a service that converts audio & video files
- Chrome/Firefox extension libraries, which count because I say so
- and today, Google Maps

This is mostly because the head developer/founder spends most of his time working on support requests/sys admin/customer stuff, and the other programmer spends ALL of his time working on the enterprise-y part of our app, so I'm left to pick up all the remaining odds and ends.

At the very least, I'll never be able to say my job is repetitive (unless reading piles of API docs is repetitive)

Also increases the chances that you'll have a brilliant mashup idea that makes you a billionair some day and it's pretty good to have a growing list of APIs you worked with for your CV especially if you can expect them to stay relevant for a bit in case you are ever looking for a new job

Has anyone written any android apps?

I may have to but I'm putting it off.

I hate Java, oh wait, not Java, but whatever it is they're calling their Java clone.

turns out something was being finicky about what order you added child controls to Control collections when re-laying things out

but only on windows xp

my "random walk code changes" bug fixing technique strikes again :P

I tried to psychic debug it for fun, here's what I came up with:


I probably shouldn't try to guess this from an offhand comment, but..

Let's call the window you have trouble with hWndBad.

Since you are in a tab control, is hWndBad visible (style) and not covered (z-order) when the invalidation occurs?

Does hWndBad get paint messages? Spyxx will tell you this conclusively.

Does hWndBad get invalidated? Test this: At some time when you think it should be invalid, call GetUpdateRect/Rgn, or try PeekMessage(hWndBad, WM_PAINT, WM_PAINT).

If hWndBad does get invalidated, and your paint code executed, but it looks wrong, that says maybe look at WS_CLIPSIBLINGS or WS_CLIPCHILDREN of other windows.

If hWndBad does get invalidated, and your paint code did not execute, somebody else validated the window...

If hWndBad doesn't get invalidated, definitely look carefully at visibility / zorder.

If you do an InvalidateRect (or RedrawWindow, if you like fps) of the tab control in code, does that change things?

A really slick way to find out how much of the window is exposed I thought of:

InvalidateRect(hwnd)
hdc= BeginPaint(hwnd)
hrgn= GetRandomRgn(hdc)
Trace(GetRgnBox(hrgn))

---

The 'traditional' way is with region math. You GetWindowRgn(GetDesktopWindow), then GetWindowRgn(EnumWindows), taking pieces out of the desktop until you get to the target top level window. Then do the same at the child level. There are a bunch of gotchas and corner cases that way though.

I did actually look at it with Spy but there was so much crap to wade through, I gave up and just fiddled with the .NET code

and that worked so now I can't be bothered to try it again

I'm pretty sure it was acting exactly as if SuspendUpdates() was called and never resumed, although I'm not sure exactly which messages that translates into off the top of my head. Not WM_SETREDRAW anyway since I tried that one

Yeah, with spy you want to filter down to a particular window + children, and just one or two messages. Click "save as default" every time...

Since there is Z-order stuff going on, maybe SetWindowPos with SWP_NOREDRAW?

Did you ever starve while dining with philosophers?

No but my donkey did get caught between two equidistant haystacks of equal deliciousness.

haha

Yeah I kind of want to write some bridge apps for android. There is nothing at all out there of any value as far as I can tell.

I've been spending the past few months doing it. Some of the limitations =


I've had to rewrite almost every app after hitting a point where I couldn't move forward with how I originally wrote it

But, for the most part you can google whatever you need to know which is nice

Should be hay and water imo and it's obviously not an Austrian donkey

well played

Here's something for the more math-y programmers in POG

I have a Google map. It has a variable number of points that are to be marked on it. The points are determined before the map is displayed. I need the map to show all the points without scrolling or zooming, but without having an overly large viewport

So, given the lat/long of each point, any suggestions on how to determine:

a) The centre of all those points
b) The required viewport size
?

zurv,

are the box boundaries true lat/long lines for all intents and purposes?

No. It's a zoom level, which is essentially meaningless outside of a google map

I think I can figure out the boundary lines for long/lat for a given physical map size and zoom level

EDIT: Zoom level = a scale, and the map is a static size on screen, so scale + size tells me how big the map will be