Fr0sT.Brutal

Sure! I'm not insisting in any form on my layout because I guess it's too personal. But storing big packages inside each project is not an option for me either (imagine 10 copies of JVCL...). It's not just a question of space - I perform "search in files in project dir" quite often and the more stuff is laying inside, the more irrelevant results will appear. Moreover, whole system of BPLs suppose global installation. Probably you guys authors of all PMs should communicate and establish a single standard of config so that PM1 could utilize PM2-prepared package. This way you won't split the (currently) small % of projects and people using PMs into several parts but unite them. To conclude: as for me, personally, what I'd use PM for: - Management of global packages - backup from existing IDE instance, batch installation to a new one with all the library/search path stuff - Packing of a project with ALL deps, including global packages

To be fair, VSCode has 5k+ open issues 🙂

A custom wizard for this purpose must be pretty easy to implement 😉 or it probably could be added as a feature to expert monsters like GExperts or CnWizards

I'm looking at these tools from time to time but still haven't adopted any of them. One of the reasons is my approach: - 3rd party packages are global (in Coding\Components) - Packages and projects I modified or develop are global in separate folder (in Coding\Projects) for small projects that unlikely need consistent history - Packages and projects I modified or develop are Git submodules inside big projects - this way I have exact pointers to repo of used packages. I doubt any PM could handle all of this. Anyway updating a package is just a question of Git pull + package rebuild so not a big mess.

How about performance?

Probably if you use CVS features of the IDE itself. I've never used it so can't say more precisely

Alas, no. The tick rollover could happen at any time, so even short-running apps are not protected. Moreover, the issue isn't exclusive for 24/7 servers only because of hibernation. I personally prefer to hibernate my PCs (both work and personal) when I leave so I have uptimes counting weeks and months.

"No answer" is a half of a trouble. Much worse is "Nvm, solved"

function TickDiff(StartTick, EndTick: Cardinal): Cardinal; begin if EndTick >= StartTick then Result := EndTick - StartTick else Result := High(Cardinal) - StartTick + EndTick; end;

They made GetTickCount64 a method of TThread *facepalm*

I personally don't see anything useful I could add without actual code.

I have 1920*1080 sec and 1920*1200 pri. IDK how to make per-monitor scale

If your small items are record or static arrays, the sort has to copy memory for each pair of them. If you use System.Generics.Collections.TArray.QuickSort<T>, it calls comparer function several times and performs 1..4 memory copying in each iteration. Anyway, instead of guessing, minimal code sample would be much more useful

So you have 2 apps and intermediate file when you already have network server all app instances interact with. Why multiply matters? OnTimer: resp := HTTPReq('server/maintenance') if resp.Code = 200 then MsgBox('Server shutdown scheduled at ' + resp.Headers['X-MaintTime'])

+1. I was expecting BtoA(AtoB(A)) == A for any A

AFAIU he's using large arrays of small arrays @Jud I think you better prepare simplified test project and show it here. You can use Github's gists to be able to update the text.

JFYI - VSCode seems using this project, maybe you find something useful in there

Is it really so important to disallow instance creation?

Can of worms, as for me. I implemented piping of console processes and created some kind of limited console and had too much mess with quirks. Output buffering, direct console drawing, plenty of escape codes and so on.

2015-03-11T23:04:42.837Z

Mongo, Redis?

"Current directory" is a per-process property that could be set to anything when a process is launched and later changed to anything by the process itself. By default (if not set by the CreateProcess parameters) it is inherited by launched processes. So. If your "Photoshop" really changes CDir to the path of currently opened file (which is weird IMHO) and doesn't specify CDir in the CreateProcess call, you can just invoke GetCurrentDirectory and chill. If it doesn't, all you've got to do is to suppose things with non-100% probability and dive into the abyss of dirty hacks. F.i., you can list handles of files opened by the parent process and find the full path but this requires elevated privileges and there's a chance that two files with the same name are opened at the same time.

So the issue is looping through data not allocation. At the 1st glance, if you could calculate the matching criteria beforehand, you could build a structure that ties criteria and item index so that at the next step you compare only flags/short hashes to receive indices.

Just an idea... is it possible to inject a console window itself into a dockable panel? Maybe it would be easier than full terminal emulation which would lack many features anyway.

You mean, launching bat file instead of your project? How it could help? There's no principal difference between running cmd.exe and yourproject.exe