Fr0sT.Brutal

I guess this very interesting subject should be split into a new thread 😉

Usually if some lib asserts itself to be fast, it must at least include benchmark results and ideally the reproducible tests. Otherwise this label is not more trustful and reliable than "the best in the world" assertion.

Yep but I guess it's considered obsolete.

O_o so you can send SMS as anybody, for example, "Bank of America" or "Interpol"?

XPman was there in D7 days, now the manifest is included by Project options -> Use runtime themes

+1. I'd consider SD's a very unreliable storage (especially micro SD's - the smaller the more fragile they are).

Dude probably you should start a blog instead

I use Git from cmd and never call gitbash so no problem.

Changed all Mac stuff in Posix.Messages to stubs (result is attached as well), fixed some more things (Longint etc). So far 52 ICS units could be built and 54 couldn't 0001-some-more-changes-to-build-for-Linux.patch Ics.Posix.Messages.pas

In 10.3 they implemented it twice - in System (as private) and in Classes as TThread's static method. I'm not sure which one of these is the most stupid *facepalm*

Or to copypaste its contents to avoid including whole Classes for this only purpose

Fixes to current SVN trunk to build under Linux 1) Change "stream.Seek(N, soFromBeginning)" to "stream.Position := N"; add some typecasts to Longint to avoid ambiguous overload call 2) Some fixes to build - commented out fragments still must be implemented! (I haven't found where 'tms' and 'times' are declared). 0001-Change-stream.Seek-N-soFromBeginning-to-stream.Posit.patch 0002-Some-fixes-to-build-under-Linux-still-must-be-implem.patch Posix.Messages still seems the biggest task

No I figured it out already. There was missing IFDEF. The main problem is Ics.Posix.Messages as it uses plenty of Macos types and functions.

Hmm, tried building ICS nightly, compiler complains about overloaded functions not marked with directive in impl section though they're marked such in decl section. Very weird. D10.3 upd Got it, some forgotten directives

You can use SemVer versioning scheme to determine breaking changes and parse project/package files to determine dependency tree so you'll get a list of packages that need to be updated.

There is libgit and https://github.com/libgit2/GitForDelphi but calling exe will be easier indeed

For such basic features as font/color styling TRichEdit is quite enough. Changing style of text fragments is several lines of code

Hmm, VirtualBox? It's free and even open-source

I have no experience with weak refs. Could you describe possible issues?

That FAQ covers this case. They say this combination violates GPL indeed. But as plugin developer doesn't own main app, no troubles - he turns out to be the violator. Anyway I believe the license regulates distribution not end user config so you on your machine can use whatever combinations you like. But I could be mistaking here.

I found this https://github.com/madorin/fibplus Starting from D2009, string contains 2-byte widechars. Depending on code, transition could take from minutes to months (if many code relies on assumption that Char is 1-byte sized). There was no critical compatibility breaks in dialect 3. Again, depending on code and complexity of queries, it could take from simple structure clone and data pumping to long and hard process. Again it depends on how hard your code is tied to data access components. In theory, changing them is a matter of several hours. Nowadays I'd advice to use a maintained library like *DAC or Zeos

It's more strict. Nevermind, I forgot GPL is TOO viral. It's LGPL I was thinking about (LGPL) Nope. Any linking is just like using the code. link

It's enough to use Move for internal relocations (of course, with zeroing an old copy - important nuance) and use assignment in other cases. I do this in my TRecordList<T>

I use MS compiler with no issues because it supports UTF8. At least XE2's brcc doesn't. And it supports more control styles and properties.

Probably for the same reason why "function MyFunction(): array of Integer;" is not allowed