-
Content Count
1940 -
Joined
-
Last visited
-
Days Won
25
Everything posted by Attila Kovacs
-
To which Delphi? Since XE2 there is System.Zip.TZipFile which would be the right way to move to.
-
There are no issues with D2007 on 21H2 at all. I have never seen an app complaining for a file which is on the location it's being searched and the app has the rights to using it.
-
FYI: Old security BUG fixed from ZLib
Attila Kovacs replied to Tommi Prami's topic in Tips / Blogs / Tutorials / Videos
Was the sechole bad? Can you/someone generate the objs? -
do you have any technical explanation for that or are you telling stories here?
-
There is more with these scrolling components. Touch / Pan is not ok as windows translates wm_gesture/pan messages to wm_scroll if the previous pan is still running.... Also, the pan handling in the VCL is wrong https://quality.embarcadero.com/browse/RSP-37027
-
"Accepted safety standards" no such thing. He can sue one for many other reasons. Insure yourself.
-
Why isn't this dangerous, or is it?
Attila Kovacs replied to Renate Schaaf's topic in Algorithms, Data Structures and Class Design
Well, it was the last option 😉 As the example shows, this pattern hiding something from you. It's not your friend. Do not underestimate that. With time, you won't remember everything. I mean, a thing. You won't remember anything. 😉 -
Why isn't this dangerous, or is it?
Attila Kovacs replied to Renate Schaaf's topic in Algorithms, Data Structures and Class Design
After the "Free" the reference is a "dangling pointer". The ShowMessage could still work for centuries but only by luck as the memory where the class was is now free for other things. So no, its not safe. Try to use explicit Free's and you don't have to keep the implementation details in your head. Or keep it there and follow the rules. -
Is Move the fastest way to copy memory?
Attila Kovacs replied to dummzeuch's topic in RTL and Delphi Object Pascal
that's how it should work actually -
it's double encoded "=" = "%3D" "%3D" = "%253D"
-
Strange Benchmark Results; Am I Missing Something?
Attila Kovacs replied to Joseph MItzen's topic in I made this
Nice. Added the x64 asm, enhanced the chunk size calculation, reduced from 8 logical to 4 physical cores (this improves a bit in 64 bit mode / under a given runtime? / on my cpu?). edit: in the IDE 4 is faster, outside the ide 8 is much faster It's now almost twice as fast as my single threaded asm was. So there is a lot of overhead firing up the threads, but still a performance gain. Project1.dpr -
Strange Benchmark Results; Am I Missing Something?
Attila Kovacs replied to Joseph MItzen's topic in I made this
Ah I see. Also, I missed one 0 in the loop, so the asm above takes ~1 sec on my pc not 100msecs. -
Strange Benchmark Results; Am I Missing Something?
Attila Kovacs replied to Joseph MItzen's topic in I made this
a good compiler would do something similar: function mod35s: int64; label l1, l2; asm push rdi push rbx mov edi, 1000000000 xor rax, rax mov rbx, 0 mov ecx, 0 @l1: imul edx, edi, -858993459 mov ecx, edi cmp edx, 858993459 jbe @L2 imul edx, edi, -1431655765 cmp edx, 1431655765 cmova ecx, ebx @L2: add rax, rcx dec edi jnz @l1 pop rbx pop rdi ret end; -
Micro optimization: IN vs OR vs CASE
Attila Kovacs replied to Mike Torrettinni's topic in Algorithms, Data Structures and Class Design
You are mostly benchmarking stack operations, calls etc.. -again- everything which has nothing to do with the actual comparisons and it's asm implementations. I'd suggest you to buy an assembly book first and try to understand the asm in the Delphi debugger. There are 3 changes in the unit, try to find them and tell us what do you think about the new results. Project1.dpr -
Experience/opinions on FastMM5
Attila Kovacs replied to Leif Uneus's topic in RTL and Delphi Object Pascal
I have an app which downloads table data's from a server with CleverComponents + OmniThreadLibrary. Data is json, zipped, encoded. All CPU is used for downloading and importing the data into an SQL server. Here are some results with FastMM 4 (default) vs. 5 using Berlin U2. 32bit FastMM4 Sync done in 138,29 Sec 32bit FastMM5 Sync done in 106,903 Sec 64bit FastMM4 Sync done in 144,357 Sec 64bit FastMM5 Sync done in 107,587 Sec I was shocked by these numbers so I thought I'm sharing my experience. -
I'd bet your program saved it at the end without the full path, so you just have copied the right one to the right place ?
-
Or you can run it as a standalone application as well, sharing the units.
-
I don't think that attaching 3rd party processes is the official way to debug isapi or other services. Is it?
-
Delphi 11.1 is available
Attila Kovacs replied to Uwe Raabe's topic in Tips / Blogs / Tutorials / Videos
for sure not just because of the numbers -
Analyze strings for common parts
Attila Kovacs replied to Mike Torrettinni's topic in Algorithms, Data Structures and Class Design
Maybe you could hire a programmer to make that, what is your budget? -
How to get ThisFormClass (TForm1) from Form1
Attila Kovacs replied to Qasim Shahzad's topic in Algorithms, Data Structures and Class Design
This is pretty true for everything in life. This is all made up, there is no need to mix anything, there should be rules to follow and that's it. I'd also find the problem in a very short time. Also, where would you store all the opened non-modal forms and why would you do that in first place? Let me guess, are those "big, complex projects" which are full of cross-form references? Form2.Edit.Text := 'lala'?- 20 replies
-
- form creation
- refactoring
-
(and 3 more)
Tagged with:
-
Haha, thx Dany, I have to admit I'm really busy with other things these days and also, his avatar mesmerizes me all the time I'm seeing it. But I'll keep an eye on him from now on 😄
-
How to get ThisFormClass (TForm1) from Form1
Attila Kovacs replied to Qasim Shahzad's topic in Algorithms, Data Structures and Class Design
there is no problem with that pattern and there is no simpler pattern and it introduces no unnecessary dependencies- 20 replies
-
- form creation
- refactoring
-
(and 3 more)
Tagged with:
-
Ahh, it's you again, emailx45 the warez king and comic book flooder. <o>