[offtopic] FreePascal

[srvaldez]

@dodicat
here's a 7z archive of fpc 3.04
please check for viruses, I think my PC is clean but you can never be too careful.
almost forgot, it's 64-bit version
let me know when you have downloaded the file so I can remove the link.

[dodicat]

Thanks srvaldez.
I'll try that later.
The pascal code runs well on virtual pascal, with only two minor adjustments.
I copied the vp21 folder directly from my xP machine to Win 10.

I get about 15 seconds in virtual pascal.

[dodicat]

OK srvaldez.
I have your file.

-0.169075164
-0.169059907
10.42882303687671 seconds

I'll fix up my dev-pascal ide to use it.
Thanks again.

[srvaldez]

you are very welcome

[jj2007]

my times were about 11 seconds on both the FP and FB

There shouldn't be a big difference because this test is full of square root calculations - and that is an extremely slow instruction. So it's all hardware dependent.

[marcov]

I get about 10 seconds with -gen gcc -Wc -O3
I cannot run the risk of using a .exe installer on win 10. (

Why not? It is a proper windows installer. Inno setup.

Does it write to the registry?

At the very least the location of the uninstaller and other settings to make a program visible in the "installed programs" dialog. It can also modify the path, but there is a tick box to not do it.

does it put files all over windows?)

The installer engine is up to date, and as of windows vista and later, stuff is rarely thrown in windows dirs anymore. If so, it is added to "common files" and the uninstaller removes it. FPC does not anyway.

I know it is free, but any half decent compiler should have the option of a .zip.

I don't see why it is different from any other Windows application. Maybe in time it will be a .msi though.

[dodicat]

Thank you marcov.
svarldez's zip fpc works well here.
Windows:
I note that If I had used the installer and agreed to put fpc.exe on path then I would also have put a 67 KB gcc.exe on path, along with a multitude of other executable files.
What then If I decide to download MinGW and have a gcc proper on path (1.5 mb approx) ?
Linux;
I have the gcc development package installed on my Linux box.
What If I tried installing freepascal on Linux, how would that affect my Linux gcc?

An ide that requires it's compiler on path (geany I believe, scite , notepad2, codeblocks (not sure about all of these), I avoid.
dev-pas ide is OK, and of course fbide, and the other FreebasiC ides seem OK.

moral:
This is why I no longer install by .exe installers.
Especially Win 10 where the folder program files is unusable, and there is no installation cd/dvd.

[marcov]

Thank you marcov.
svarldez's zip fpc works well here.
Windows:
I note that If I had used the installer and agreed to put fpc.exe on path then I would also have put a 67 KB gcc.exe on path, along with a multitude of other executable files.

Yes, that's why it is an OPTION. And even if you did, you can simply edit it. Nearly all toolchains (even Visual Studio or Delphi) pack an own build or version of at least MAKE, so that is normal anyway.

[dodicat]

Fair enough marcov.
But I still think that you should have a portable(zip, 7Z,whatever) version of FreePascal.
Other compilers have, Dev-C++, FreeBASIC of course, Haskell, ...
You might attract more punters (sorry ... developers).

[marcov]

Fair enough marcov.
But I still think that you should have a portable(zip, 7Z,whatever) version of FreePascal.
Other compilers have, Dev-C++, FreeBASIC of course, Haskell, ...
You might attract more punters (sorry ... developers).

We had it. It had 5000:1 download ratio. It is just not worth the trouble for the kind of users that have been shown to give up easily anyway.

[badidea]

An ide that requires it's compiler on path (geany I believe, scite , notepad2, codeblocks (not sure about all of these), I avoid.

About Geany, you can just tell it where your fbc.exe, python.exe, fpc.exe, etc is. Last one never tried.

[marcov]

FPC 1.0.6 is old (2002 or so, but from a branch that was already in stabilization for 1.0 before 2000) , and only had a simple register colourer as optimizer, which only functioned half because of certain constraints.

FPC 2 are nearly complete rewrite, and not comparable. So benchmarking with 1.0.6 is a waste of time. Yours, and ours.

[jj2007]

FPC 2 are nearly complete rewrite, and not comparable. So benchmarking with 1.0.6 is a waste of time. Yours, and ours.

My benchmarks here use version 3.0.4, so I hope they are relevant.

[marcov]

FPC 2 are nearly complete rewrite, and not comparable. So benchmarking with 1.0.6 is a waste of time. Yours, and ours.

My benchmarks here use version 3.0.4, so I hope they are relevant.

The textual read is not what I expect, but I don't have time to investigate. Things changed there because of unicode support.

I see that my reply was for an older message in the thread by dodicat, I accidentally went to the end of the thread using "page 2" instead of "page 3".

[dodicat]

Indeed I did test with a very old freepascal because that was all I had.
I also tested with virtual pascal, which actually was quite fast (15 seconds).
(Virtual pascal has been dead since 2005)
But I now have freepascal 64 bit version 3.0.4 thanks to srvaldez.(10 seconds)
Anyway, I am a little disappointed that -gen gas was so slow at this planetary benchmark (23 seconds).
I tried using (somenumber*somenumber) instead of (somenumber^2) but it made hardly any difference.