VBA Rubberducking

^ that'll be part one. I had to stop as I was getting cross-eyed reading the text. Got as far as the EmptyUIRefresh.xaml

Cool, thanks!

:+1:

@KySoto meh... pretty much like every other language. Once you get some practice, it's not too difficult. Just be sure to forget everything you know about OOP. :) I'm pretty sure newer tools are decent at helping you write structured code, though.

I know for sure I wouldn't want to try and write Rubberduck in assembly.

2

8:24 PM

probably not many desktop apps that should be written in assembler. There are good uses for it, and I learned a lot about programming in general by learning some assembler.

@FreeMan I don't think many are. It's mostly C.

And C++.

Definitely. I can see why you'd want to embed a bit tiny piece of assembly code in some hot path.

but to write a program in it alone?

even something simple like a calculator app would be very very tedious to write in assembler.

I'm a bit disappointed that there isn't an RD instruction in x86 assembly.

5

It's good for embedded programming and real-time processing. And it makes a great teaching tool.

@Comintern lol

8:27 PM

You don't have to know how an internal combustion engine works to drive a car, but it gives you a better feel for what's going on around you.

yep

Stupid question --- why "real-time processing" specifically? Are C/C++/etc. not "real-time"?

@this They can be.

That's my point - if they can be, then why need to sink down to assembly then?

But you can still do better than them if you are excellent.

ah i see.

8:30 PM

You can use less space, you can be faster, etc. if you program for a specific use-case or device.

I'm guessing that the odd of a human outsmarting the compiler's optimizations is quite low, though.

Because a compiler is heinously smart, but it has to compile for the generic case.

@this Yep.

yeah

You basically have to be good enough to be able to write a compiler.

And there's a reason most real-time is in C or C++.

But you can write a problematic snippet in assembly.

and you must be awesome if you can write an OS in VBA!

8:32 PM

Pfff, that's easy, if you want something as slow and single-minded as a punchcard machine.

lol

@this Don't give Mug any ideas. He's had a few already.

uh, I definitely don't claim credit there

Hmmm... I wonder if you could build an instruction pipeline out of Excel formulas.

@Comintern You should know better...

8:34 PM

@IvenBach between me and @Comintern, I'm not the dangerous one ;-)

Yes, yes I should.

The interesting thing about an Excel spreadsheet instruction pipeline would be that you could create something similar to self-modifying hardware.

^ see, proof. give the man enough time, he'll turn Excel into Skynet

Just tried rebuilding after pulling from upstream/next. Still have build failures.

One of these days that magic build will happen.

Is "My spreadsheet in sentient, how do I kill it" on topic on SO, or is that more of a Super User question?

2

I'd go with WorldBuilding

8:40 PM

Step 1. Pull the plug on the internet.

Step 2. Pull the plug on the computer.

Step 3. Remove the battery from the computer.

Step 4. Destroy the RAM and HDD/SSD.

Step 5. If you didn't do step 1 fast enough, fight angry terminators who are pissed that you killed their daddy/mommy.

@IvenBach i think I told you before - kick your motherboard and pour some coffee in it

then tell your boss you need a new computer.

Also... exactly same errors? Nothing new?

I'd give it another shot after the CE changes merge.

They appear to be the same messages still.

Oh ****.

I just realized why China is the world's recycler.

What a goldmine of information that is!

Lack of environmental regulation?

8:53 PM

They can rummage through that for all sorts of crap. Secret information, clothing styles, newspapers, maps, et al.

@this lol

M.Doerner

9:14 PM

Wasn't there already some project that built a programming language for circuits simulated on Excel spreadsheets?

Ah, no. That was for that game of life thing.

in The Factory Floor, 26 secs ago, by Gryphon

@Hosch250 That's... actually the correct place for that question, if its properly constrained, etc.

M.Doerner

However, I think you can simulate that on Excel spreadsheets.

@M.Doerner I've seen full-on games implemented with Excel spreadsheets.

Super-mario, 2048, and more.

Some of these didn't even use any VBA.

M.Doerner

Using VBA would not really count.

Wow, hating on DataGrid is apparently a thing.

> The 15 yr old Delphi datagrid runs faster on 15 yr old hardware.

> Datagrids really highlight the weaknesses of WPF.

9:27 PM

been fighting with one all day now

if I can get the newly added row to scroll into view when the "create" command runs, then it breaks normal scrolling and keeps the selected row into view

I was trying to figure out last night if I could style a ListView to replace it, but that has its own problems with the column widths.

it's maddening

Hmmm... I wonder if that's the same issue that the non-WPF DataGrid has - the event order of determining the height of the paged window is weird IIR.

I'm trying to remember what I did to hack around it.

WPF: making braindead-simple CRUD apps challenging again since 2006

IKR?

OLE embed an Excel spreadsheet.

Call it good.

Ah yes, same problem in the WPF version:

> It came down to an issue (bug?) with the WPF sizing calculations. I had it in a grid with the RowDefinition Height="Auto" which was causing the rendering system to try and recalculate the size of the DataGrid at runtime by measuring the size of each and every column and row, presumably by filling the whole grid (as I understand it). It is supposed to handle this intelligently somehow but in this case it was not.

I wonder if RD's is doing that. I should test...

9:40 PM

well, that black hole of memory leak didn't just appear magically out of nowhere....

I was also thinking about that, and wondered if replacing an actively bound ObservableCollection<T> at runtime can leak it.

hmm. i know that for each parse run, we increase 20 MB

i.e. Bound = new ObservableCollection<foo>(); instead of clearing it and re-adding members.

which is a small fraction of the entire memory leak

IIRC, we get about 200 MB increase after a parse or somethin glike that.

not surprising

9:45 PM

The point is that if we're leaking because we're replacing, it should be leaking 200 MB, not 20 MB, right?

IIRC the CE VM was leaking the whole parser state... that would be pretty much all of that 200MB

How much of that is eligible for GC though?

That's the other thing - 20 MB will eventually get freed.

so it's not really a "leak"

a true leak would never get freed.

RPS is just working memory. that we leak it is a different issue than it being 200MB..

Wasn't the parser state leak due to simply not unsubscribing an event?

9:47 PM

no idea

that chat was well over a year ago IIRC

I thought that was already fixed in one of past PRs.

but the memory footprint is still as big

I'll need to check my diff. I do remember fixing a potentially substantial leak.

btw, i've seen it claimed that CE VM leaks RPS -- but how do we know that exactly?

I could have been any number of things TBH.

you?

9:51 PM

(might still be)...

@this GH has two (or maybe three??) different ways of rendering diff

you mean split vs unified?

not sure that helps.

honestly, I very much prefer interspersed +- lines to git / GH just outright giving up and rendering the stuff as two different files, one in green, one in red

when it's changes in different places, that makes perfect sense

but when it's one big block of changes, it makes no sense to intersperse....

(within that one block, I mean)

that whitespace thing also is incredibly helpful

usually eliminates tons of diff for small refactorings

9:57 PM

hmmm.

tried it doesn't really work out

here's an atrcious example where I think it should be one block:

github.com/rubberduck-vba/Rubberduck/pull/4661/…

Duga

[rubberduck-vba/Rubberduck] Vogel612 deleted branch issue-templates

looking at it closer, the real problem is with the lines where there's just a { or }

that causes GH to break up the blocks

which makes it hard to follow the changes when they technically are one big blocks.

probably causes git to do that, actually..

git might actually be able to be a bit smarter about this with --minimal, but that costs extra compute and GH won't spend that compute if they can help it

perhaps. But it's annoying still.

yeah. for now, View files in a 2nd tab that comintern suggested works well enough for comprehending the functionality of the code

@this that does look better in split view

10:10 PM

as long the blocks are smaller, yeah

for big blocks, you get a big gap.

such is life, can't have it all

like, can't have a DataGrid and a working app, apparently.

oh damn, looks like I need to implement IEditableObject on all viewmodels

fml

wait, so they were lying when they said to collect them all?!?

it's marketing, they're always lying

"Collect all 6!" - has one in a million chance of getting the 6th one

aw, that reminds me of that old marketing thing they had with cokes.

used to have prizes under the caps

more likely, you were to get a free bottle. I don't think anyone actually won the advertised prizes (whatever it was)

fml, ttqw

Duga

10:28 PM

[rubberduck-vba/Rubberduck] bclothier commented on Rubberduck.SettingsProvider/XmlPersistanceService.cs of pull request #4699: Clean up the persistence services

[rubberduck-vba/Rubberduck] comintern commented on Rubberduck.SettingsProvider/XmlPersistanceService.cs of pull request #4699: Clean up the persistence services

@this it's always some old lady that has nothing better to do than to fill up thousands of mail-in coupons ("no purchase necessary")

or the CEO's step-son

11:01 PM

RD code snippets are already on the horizon?

If not #MakeItSo.

The more I learn about VS and tools it lets you do the more I wish I worked with it.

> Excel, baby... I think you're no longer my favorite.

KySoto

@FreeMan i started out on structured code, the first time, and the second time. third time, i did OOP

though, a lot of people did try to add OOP concepts to it.

first time was logo, second was Jass scripting for warcraft 3

third was when i went to school

hmm I don't know if I like this:

container.Register(Component.For(typeof(IEnumerable<Type>))
    .DependsOn(Dependency.OnComponent<ViewModelBase, GeneralSettingsViewModel>())
    .LifestyleSingleton()
    .Instance(experimentalTypes));

why not create a ExperimentalTypesProvider?

IEnumerable<Type>, even restricted to only the settings viewmodel seems a bit too fast and loose.