VBA Rubberducking

mansellan

12:00 AM

yeah same error on a fresh direct clone :-(

Duga

RELOAD!

[bruglesco/memory-card] 6 commits. 365 additions. 303 deletions.

[Cardshifter/Cardshifter] 1 commit. 1 opened issue. 1 addition. 2 deletions.

[Cardshifter/HTML-Client] 11 commits. 7 opened issues. 6 closed issues. 11 issue comments. 451 additions. 331 deletions.

[FreezePhoenix/XtraUtils] 2 commits. 2 additions. 1 deletion.

[rubberduck-vba/Rubberduck] 2 opened issues. 1 closed issue. 13 issue comments.

[Zomis/Server2] 1 opened issue.

M.Doerner

12:23 AM

At the start of this year I had been intending to try to create a COM shim for RD. However, then the point was made that we should get the leaking completely objects under control before masking the problem with a shim.

Moreover, introducing a COM shim runs into a few problems.

First of all, we would need two of them, one for x86 and one for x64.

AFAIK, there is no Any CPU for native c++ assemblies.

Moreover, we would run into issues with security permissions.

mansellan

@M.Doerner Makes sense, there's no VM in the middle

@M.Doerner Why so?

M.Doerner

Currently, we are side-stepping office security settings because these check the entry level dll of an add-in. For us that is the one of the CLR because we are hosted in it.

With a shim, the check is against the shim.

mansellan

@M.Doerner Ah yes... so we'd get into needing certs & signing...

MZ must have solved for it though?

M.Doerner

Being enabled by default under high security settings and allowed at all under very high settings requires a strongly named assembly.

MZ is no longer OSS. So, managing the key for the strong naming is not a problem.

mansellan

Hmm yeah...

Mathieu Guindon

12:31 AM

@Vogel612 not that I know of

M.Doerner

It is not very OSS friendly.

Mathieu Guindon

So, screw it

mansellan

^^

M.Doerner

Btw, the latest vs in which the shim wizard works is VS2010.

Mathieu Guindon

Yup

mansellan

12:33 AM

Nice!

M.Doerner

It is not as bad as it sounds, though. You just need it to generate the files. It should all compile in later versions.

mansellan

What's the betting it fails to work with 2010 Express too?

Just for my understanding - is key management for OSS a showstopper or just a PITA?

M.Doerner

12:54 AM

I think it is primarily a pita. This is not made better by the need to sign all assemblies if you sign the entry point.

Mathieu Guindon

@mansellan what's the point of a private key in a public repo?

(unless I'm missing something)

M.Doerner

To satisfy the requirement of other people to have a strong name on the assembly.

mansellan

@MathieuGuindon more likely to be me that's missing something... but there's other secrets held by the project, is this not the same? I mean, it's only the green releases that need to be signed IIUC, and that could be done by the project owner or senior contibuter outside of CI?

M.Doerner

It is I tended at a security measure but in the end it does not provide any.

Mathieu Guindon

so, we just share the private key and call it "strongly named"

Oh, just sign the green releases

mansellan

12:58 AM

well, share among a few, not put it in GH...

Mathieu Guindon

Got it

mansellan

yeah

Mathieu Guindon

So, back to manual releases then

mansellan

but we only ship a few times a year...

hotfixes aside

Mathieu Guindon

true

mansellan

1:02 AM

just thinking that, if I've understood the docs correctly, our current setup doesn't allow us to spawn extra AppDomains, which may be needed if\when we implement a plugin feature.

anyway just my 2c, I'm happy either way :-)

Peter M Taylor

2:22 AM

To follow your thoughts in a different way for my comprehension @mansellan to guarantee a “sandbox” environment that plugins are tested separately/independently while our build improves as we have our build workflow thus need we have addition paths added into the build with AV down the track when ready to release them?

Interesting challenging hurdle when/if we get there...

Quack Exchange

4:09 AM

-1

Q: Speed up script

I have the following script which works great. The only thing is it takes a lot of time to run on a worksheet with 2000+ rows. Anyone know of a way to speed it up? The code runs through the workbook and ignores the pages I do not want it to touch. Then, it runs through any pages that I want it t...

mansellan

9:55 AM

Hmm... looks like it may be possible to have AV do the signing without exposing the key:

help.appveyor.com/discussions/questions/…

@PeterMTaylor Sandboxing is correct - for plugins we'd want to provide some level of isolation such that problems with a plugin can't take down the rest of RD (or other unshimmed addins)

M.Doerner

10:33 AM

Before we could go this route, we would have to check one thing: do we reference unsigned packages?

mansellan

good point...

or... chat.stackexchange.com/transcript/message/44820193#44820193 :-)

sounds like it could be [very] irksome

/aside: How broken is the MS security model that you basically have 2 options: 1) Jump through an insane number of hoops or 2) sidestep the whole thing by just using the default appdomain...

mansellan

10:58 AM

@M.Doerner lol there this, but, uh, no... stackoverflow.com/questions/1379954/…

fwiw I now think shimming is a non-starter

#TooManyProblems

Vogel612

We should probably document that somewhere in an issue or seomthing

mansellan

agreed. probably needs to go on the wiki i guess. for now I can add some comments to the issue Carlos commented on?

Vogel612

11:19 AM

I'd personally not search the wiki for info about the COM shim

I think an issue with a "proposal" and a "response" would make more sense..

but maybe that's just me

mansellan

ok will do

Duga

12:03 PM

[rubberduck-vba/Rubberduck] mansellan opened issue #4041: Discussion: COM Shims

> Currently, Rubberduck loads directly into the default AppDomain provided by the host application Add-In manager. Microsoft recommends against this approach, advocating instead for the use of a COM shim. This is a thin proxy, written in native code (C++) which registers itself as an Add-In and when loaded initialises an AppDomain, loads the managed assemblies into it, then forwards calls from the host to the extensibility interfaces across to the managed assemblies. Full details on how this wor

[rubberduck-vba/Rubberduck] mansellan added label discussion to issue #4041: Discussion: COM Shims

Vogel612

Thanks for doing the leg work :)

mansellan

:-)

Peter M Taylor

1:53 PM

Good man @mansellan

Vogel612

8:23 PM

PS F:\cleme_000\Desktop\Programming\Rubberduck> git config --global alias.update
fetch --all --prune
PS F:\cleme_000\Desktop\Programming\Rubberduck> git help update
Die Datei "C:\Program Files\Git\mingw64/share/doc/git-doc/git-update.html" kann nicht gefunden werden.
PS F:\cleme_000\Desktop\Programming\Rubberduck> git update
Warning! `git update` has been deprecated;
Please use `git update-git-for-windows` instead.
Git for Windows 2.17.0.windows.1 (64bit)
Up to date
PS F:\cleme_000\Desktop\Programming\Rubberduck>

ugh... eff you as well, Powershell

@mansellan status-norepro :/

I'm on rubberduck-vba/Rubberduck/next

5db0eab0778355b6b95e5cd1f024efb356bb9122

^^ that commit hash.

builds cleanly

mansellan

9:00 PM

Hmm... odd indeed.

mansellan

9:17 PM

gonna try updating VS

Duga

9:50 PM

[rubberduck-vba/Rubberduck] Vogel612 removed label PR-Status: Review Requested from pull request #4006: Splice Off Rubberduck.Refactorings from Rubberduck.Core

[rubberduck-vba/Rubberduck] Vogel612 added label enhancement to issue #4042: Clean up IMessageBox

[rubberduck-vba/Rubberduck] Vogel612 opened issue #4042: Clean up IMessageBox

> IMessageBox was changed up a bit in #4006

While that was not the general intention of the PR, some cruft remains, namely the `bool?` parameter and return value on `Confirm` as well as the missing `caption` argument on `Message`.

The `bool?` should be replaced with a thin wrapper over the `DialogResult` members `Yes`, `No` and `Cancel`.

`Message` should get a (possibly optional?) parameter `caption`. If necessary, callsites need to be updated.

[rubberduck-vba/Rubberduck] Vogel612 added label difficulty-01-duckling to issue #4042: Clean up IMessageBox

[rubberduck-vba/Rubberduck] Vogel612 added label technical-debt to issue #4042: Clean up IMessageBox

[rubberduck-vba/Rubberduck] Vogel612 added label enhancement to issue #4043: IRefactoringDialogs should not use Windows Forms API

[rubberduck-vba/Rubberduck] Vogel612 added label difficulty-02-ducky to issue #4043: IRefactoringDialogs should not use Windows Forms API

[rubberduck-vba/Rubberduck] Vogel612 opened issue #4043: IRefactoringDialogs should not use Windows Forms API

> It was noted in #4006 that the `Windows.Forms` namespace is still in use for `IRefactoringDialog` usages.
That's not really a good solution and the result of a dialog invocation should be replaced by an appropriate type (either `bool` or the same enum that's noted in #4042)

[rubberduck-vba/Rubberduck] Vogel612 added label up-for-grabs to issue #4043: IRefactoringDialogs should not use Windows Forms API

[rubberduck-vba/Rubberduck] Vogel612 added label technical-debt to issue #4043: IRefactoringDialogs should not use Windows Forms API

[rubberduck-vba/Rubberduck] Vogel612 assigned Vogel612 to issue #4043: IRefactoringDialogs should not use Windows Forms API

[rubberduck-vba/Rubberduck] Vogel612 assigned Vogel612 to issue #4042: Clean up IMessageBox

[rubberduck-vba/Rubberduck] Vogel612 created comment on pull request #4006: Splice Off Rubberduck.Refactorings from Rubberduck.Core

> I've extracted the remaining concerns into follow-up issues and assigned them to me. Considering that there have been two tentative approvals and the respective suggestions have either been incorporated or moved into a follow up, I'm merging this PR

[rubberduck-vba/Rubberduck] Vogel612 merged pull request #4006: Splice Off Rubberduck.Refactorings from Rubberduck.Core from Vogel612/Rubberduck/project-splicing into rubberduck-vba/Rubberduck/next

[rubberduck-vba/Rubberduck] Vogel612 pushed commit ff0d71fa to next: Move UserInteraction to separate project, Stop leaking Windows Forms from IMessageBox

[rubberduck-vba/Rubberduck] Vogel612 pushed commit 8e92e3aa to next: Move Refactorings into separate project, see #3862

[rubberduck-vba/Rubberduck] Vogel612 pushed commit d5e529b3 to next: Include Rubberduck.Resources, fix remaining Compilation Problems

[rubberduck-vba/Rubberduck] Vogel612 pushed commit ca565030 to next: Fix compilation errors that VS was quite quiet about

[rubberduck-vba/Rubberduck] Vogel612 pushed commit 16b32a82 to next: Re-Fix CW Proxy types, Make Core internals visible to Castle Proxy Generation

[rubberduck-vba/Rubberduck] Vogel612 pushed commit a9205113 to next: Fix incorrectly set up unit-tests

[rubberduck-vba/Rubberduck] Vogel612 pushed commit a4f72c46 to next

Update RemoveParametersRefactoring to use a separate Removal collection

This allows segregating UI concerns from Model Concerns.
Also fixes two failing unit-tests because of incorrect messageBox setup

[rubberduck-vba/Rubberduck] Vogel612 pushed commit abe50ee1 to next: Correctly set up last two unit test failures

[rubberduck-vba/Rubberduck] Vogel612 pushed commit 3fedf3d4 to next

Incorporate Feedback,

Merge Prompt and Question on IMessageBox
Rename ConfirmYN to ConfirmYesNo
Add some left over Resource Strings

[rubberduck-vba/Rubberduck] web-flow pushed commit 960fb85e to next

Merge pull request #4006 from Vogel612/project-splicing

Splice Off Rubberduck.Refactorings from Rubberduck.Core, see #3862

[rubberduck-vba/Rubberduck] Vogel612 edited issue #3862: Modularization strikes again - Let's move some stuff around

mansellan

Updated VS (including adding C++), builds for me now

Vogel612

:+1:

Duga

[rubberduck-vba/Rubberduck] Vogel612 edited issue #3862: Modularization strikes again - Let's move some stuff around

Vogel612

^^ each time one adds a checkmark to that issue it's counted as edit ...

:/

Duga

[rubberduck-vba/Rubberduck] Vogel612 edited issue #3862: Modularization strikes again - Let's move some stuff around

[rubberduck-vba/Rubberduck] retailcoder published prerelease Rubberduck-v2.2.0.3291

[rubberduck-vba/Rubberduck] retailcoder created tag Rubberduck-v2.2.0.3291

[rubberduck-vba/Rubberduck] build for commit 960fb85e on next: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] build for commit 960fb85e on next: 53.4% (target 0%)

Peter M Taylor

10:38 PM

A open question to @all...has anyone successfully used a dark theme to see Github changes with black/dark background instead of while/light?

With their browser of their choice.

Vogel612

had a stylesheet for the stylish extension. it kinda sucked...

so I dropped the style again

Duga

[rubberduck-vba/Rubberduck] Vogel612 created pull request #4044: Follow Up to 4006 (WIP) to merge Vogel612/Rubberduck/windows-forms-purge into rubberduck-vba/Rubberduck/next

[rubberduck-vba/Rubberduck] Vogel612 removed label bug from issue #3453: Rubberduck doesn't see anything yet and submenus are disabled

[rubberduck-vba/Rubberduck] Vogel612 removed label critical from issue #3453: Rubberduck doesn't see anything yet and submenus are disabled

[rubberduck-vba/Rubberduck] Vogel612 removed label support from issue #3151: Resolver Error in build 2.1.1994

[rubberduck-vba/Rubberduck] build for commit c45bba82 on unknown branch: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] [codecov[bot]](github.com/apps/codecov) created comment on pull request #4044: Follow Up to 4006 (WIP)

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/4044?src=pr&el=h1) Report
> Merging [#4044](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/4044?src=pr&el=desc) into [next](https://codecov.io/gh/rubberduck-vba/Rubberduck/commit/960fb85ebf426feda08520fa4876adeb227056ec?src=pr&el=desc) will **decrease** coverage by `<.01%`.
> The diff coverage is `25%`.

```diff
@@ Coverage Diff @@
## next #4044 +/- ##
==============================

[rubberduck-vba/Rubberduck] build for commit c45bba82 on unknown branch: 53.4% (target 0%)

[rubberduck-vba/Rubberduck] mansellan synchronized pull request #4021: VB6 - Exclude ResFile and RelatedDocument from parse

[rubberduck-vba/Rubberduck] Vogel612 removed label support from issue #2988: AutoSave causes code pane to lose focus

[rubberduck-vba/Rubberduck] [codecov[bot]](github.com/apps/codecov) edited comment on pull request #4021: VB6 - Exclude ResFile and RelatedDocument from parse

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/4021?src=pr&el=h1) Report
> Merging [#4021](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/4021?src=pr&el=desc) into [next](https://codecov.io/gh/rubberduck-vba/Rubberduck/commit/960fb85ebf426feda08520fa4876adeb227056ec?src=pr&el=desc) will **increase** coverage by `0.38%`.
> The diff coverage is `28.57%`.

```diff
@@ Coverage Diff @@
## next #4021 +/- ##
=========================

Vogel612

11:09 PM

okay, one can search for issue creation ...

I'll head to bed for now...

Duga

[rubberduck-vba/Rubberduck] build for commit 77b71b7d on unknown branch: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] [codecov[bot]](github.com/apps/codecov) edited comment on pull request #4021: VB6 - Exclude ResFile and RelatedDocument from parse

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/4021?src=pr&el=h1) Report
> Merging [#4021](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/4021?src=pr&el=desc) into [next](https://codecov.io/gh/rubberduck-vba/Rubberduck/commit/960fb85ebf426feda08520fa4876adeb227056ec?src=pr&el=desc) will **decrease** coverage by `0.02%`.
> The diff coverage is `27.5%`.

```diff
@@ Coverage Diff @@
## next #4021 +/- ##
==========================

Vogel612

might drop some more work onto the MessageBox stuff tomorrow, but I kinda doubt it..

Duga

[rubberduck-vba/Rubberduck] build for commit 77b71b7d on unknown branch: 53.38% (target 0%)

[rubberduck-vba/Rubberduck] mansellan removed label PR-Status: Conflicting from pull request #4021: VB6 - Exclude ResFile and RelatedDocument from parse

Vogel612

Tuesday I'll be able to look into stuff, unless my Paper is being recalcitrant

mansellan

nice work @Vogel612

Transcript for