|
|
I'm in a "sit-back-with-a-hard-cider-and-some-popcorn-and-watch-the-world-burn" sort of mood today  .
Software Zen: delete this;
|
|
|
|
|
That would be nice to know what new amazing feature you are talking about. Then I may shout with you or laugh about this ! I am looking forward to get a new MacBook Pro 16" M1 Pro.
PS I am bilingual, I work with both Windows and MacOS  .
Gilles Plante
|
|
|
|
|
Menu Extras, soon to be released in the App Store (249.99$), will show the missing menu parts on your Apple Watch.
In the mean time, try rotating your device...
Luc Pattyn [My Articles]
If you can't find it on YouTube try TikTok...
|
|
|
|
|
This bit of code will take an array int[] entries and some text (string or char array or whatever) and tell you whether the text matches the state machine stored in entries
I've effectively packed a DFA state machine into an array of ints, and very efficiently unpack it while matching/walking it because of the way it's encoded. Only downside is encoding the array takes two passes, but the passes are fast.
Here _FetchNextInput(cursor); simply fetches the next UTF32 codepoint off the text input (doing surrogate pair encoding in the process when necessary)
var cursor = text.GetEnumerator();
var ch = _FetchNextInput(cursor);
if (ch == -1) return -1!=entries[0];
var state = 0;
while (ch != -1)
{
var acc = entries[state++];
var tlen = entries[state++];
var m = false;
for (var i = 0; i < tlen; ++i)
{
var tto = entries[state++];
var prlen = entries[state++];
for (var j = 0; j < prlen; ++j)
{
var tmin = entries[state++];
var tmax = entries[state++];
if (ch >= tmin && ch <= tmax)
{
ch = _FetchNextInput(cursor);
if (ch == -1) return -1!=entries[tto];
state = tto;
i = tlen;
m = true;
break;
}
}
}
if (!m) return -1!=acc;
}
return false;
Real programmers use butterflies
|
|
|
|
|
You remind me of the title of Niklaus Wirth's great book from the 1970s:
Algorithms + Data Structures = Programs
Particularly in resource-limited environments, laying your data out right can make all the difference.
(Numerous war stories omitted...)
Cheers,
Peter
Software rusts. Simon Stephenson, ca 1994. So does this signature. me, 2012
|
|
|
|
|
I came across this obituary today:
https://www.legacy.com/us/obituaries/name/mark-sanborn-obituary?id=6461229[^]
I suspect that this is the same Mark Sanborn that maintained the website where you could download a Windows hosts file to block nuisance connections:
Blocking Unwanted Connections with a Hosts File[^]
Can somebody confirm this? The owner of the hosts website has been reporting that he had severe health problems for several months now. He has not updated the hosts file since early March, which is highly unusual. If this obituary refers to the Mark of the hosts file: I appreciated his work and wish him RIP ![Rose | [Rose]](https://codeproject.global.ssl.fastly.net/script/Forums/Images/rose.gif)
Get me coffee and no one gets hurt!
|
|
|
|
|
So...life added his name to its own hosts file?
Too soon?
|
|
|
|
|
Is NJ anywhere close to Broomfield, CO?
|
|
|
|
|
Quote: Is NJ anywhere close to Broomfield, CO?
In galactic terms, yes.
- I would love to change the world, but they won’t give me the source code.
|
|
|
|
|
Jörgen Andersson wrote: Is NJ anywhere close to Broomfield, CO?
If you zoom out far enough they are yay distance apart.
|
|
|
|
|
NJ (New Jersey) is northeast USA; it touches New York City.
CO Colorado is almost center of USA.
|
|
|
|
|
Did you update your hosts file?
Start here[^].
Bastard Programmer from Hell 
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.
|
|
|
|
|
Thank you for the info about the site, I didn't know that - I used a third party software for that.
GCS d--(d-) s-/++ a C++++ U+++ P- L+@ E-- W++ N+ o+ K- w+++ O? M-- V? PS+ PE- Y+ PGP t+ 5? X R+++ tv-- b+(+++) DI+++ D++ G e++ h--- r+++ y+++* Weapons extension: ma- k++ F+2 X
|
|
|
|
|
Okay, so "code synthesis" (what I've heard it called in newsgroups and such but i don't know if there's an official term) is the process of generating code that - to put it simply - looks more like code a human would write than code that is generated.
For example:
if(ch == '\"') {
ch = _FetchNextInput(cursor);
if(ch == -1)
return false;
goto q1;
}
return false;
q1:
if((ch >= '\0' && ch <= '!') || (ch >= '#' && ch <= '[') || (ch >= ']' && ch <= 1114111)) {
ch = _FetchNextInput(cursor);
if(ch == -1)
return false;
goto q1;
}
if(ch == '\"') {
ch = _FetchNextInput(cursor);
if(ch == -1)
return true;
goto q2;
}
if(ch == '\\') {
ch = _FetchNextInput(cursor);
if(ch == -1)
return false;
goto q3;
}
return false;
q2:
return false;
q3:
if((ch >= '\0' && ch <= '!') || (ch >= '#' && ch <= '[') || (ch >= ']' && ch <= 1114111)) {
ch = _FetchNextInput(cursor);
if(ch == -1)
return false;
goto q1;
}
...
return false;
...
Forgive me for the long example, but I couldn't think of a shorter one that I had on hand that would serve. It's not the complete code, but you can see it's generated goto tables, and if you squint you can see they match text.
Humans wouldn't have written code like this - at least not typically, and without enough caffiene.
Anyway, check out just under q1:
q1:
if((ch >= '\0' && ch <= '!') || (ch >= '#' && ch <= '[') || (ch >= ']' && ch <= 1114111)) {
ch = _FetchNextInput(cursor);
if(ch == -1)
return false;
goto q1;
}
that could be
q1:
while((ch >= '\0' && ch <= '!') || (ch >= '#' && ch <= '[') || (ch >= ']' && ch <= 1114111)) {
ch = _FetchNextInput(cursor);
if(ch == -1)
return false;
}
...
And then I find other opportunities to "collapse" code by similarly examining the flow (which i already have state diagrams for in this case, but in the general case you'd build some sort of flow diagram over your generated bits like that and work with it), and then i do it over and over again until there are no more changes to the code/flow
This could yield some pretty cool results. It's nowhere near an article yet. Just an idea I had this morning while working on an update to Reggie.
Real programmers use butterflies
|
|
|
|
|
People can't be trusted to write good code; I prefer computer-generated code.
|
|
|
|
|
I know you're joking, but I agree with that sentiment. Less bugs. Code to a spec. The problem is the code is not "pretty" - this is to make it "pretty" so that once the tool no longer functions anymore the code is hand editable. =) it's next-level codegen.
Real programmers use butterflies
|
|
|
|
|
Wouldn't someone simply pull it into VS and reformat it as desired?
|
|
|
|
|
Yes, if the idea is only to have the tool help you draft the code.
No, if the idea is to evolve the code using the tool.
|
|
|
|
|
It's not about formatting, so much as it is about eliminating things like goto tables, and resynthesizing state machines into something a little more organic where possible.
Real programmers use butterflies
|
|
|
|
|
honey the codewitch wrote: Less bugs
Wouldn't that depend on who wrote the code generator?
|
|
|
|
|
It's a math problem though. If I generate code N times, 1 bug is replicated N times. If I fix the bug in 1 place, that's N less bugs.
But yes, the code should be correct. However, built correctly once, it writes correctly every time.
Real programmers use butterflies
|
|
|
|
|
Could be worse, could be O(N2)
|
|
|
|
|
One (human) generation ago, HLL compilers were introduced so that programmers wouldn't make errors confusing, say, left shift with rotational shift. Lots of coding errors were avoided.
So we moved bug creation to a higher level, such as off-by-one-errors and referencing undefined elements. Bugs became more sophisticated.
Code generation brings bug sophistication to a new and higher level. In the old days we were working with people who had a deep understanding of the problem solution. As we learned to handle that, we gradually start communicating with people who knows little beyond 'When I press this key, that happens!' So, the bugs are no longer at the level of plain vs. rotational shift, but in fundamental flaws in how to solve the problem. No one of the problem domain experts is there to oversee that we code the solution according to their professional understanding of the solution. That isn't necessary, now that no coding is required.
Or, that is what we believe. I think we need a new coding-less profession: A kind of professionals with no coding experience, who don't give as d**n about safe pointers and code templates and indentation rules, but are experts in analyzing user problems, work patterns, interaction and dependencies between different information structures. People who are professionals in defining tasks for us IT guys to solve according to the needs of the users, without being steered / dictated / limited by specific coding methodologies.
Long, long time ago I read an introduction to COBOL-60, proclaiming that the language would wipe out the need for computer programmers; now the users themselves could specify the handling of the data in this language very close to natural English language. The 60+ years since COBOL-60 was released has certainly shown us that problem is not in knowing the programming language keywords, but far more in a thorough problem analysis and understanding. That remains, whether we call the tools 'compilers' or 'code generators'.
To be honest: I cannot clearly tell the dividing line between a 'compiler' and a 'code generator'. That is doubly true if you claim that 'code generators' is something that has appeared the last 20 years or so, that earlier systems were not 'code generators'. To me, this is a lot like reinventing wool.
|
|
|
|
|
I'm afraid I can't let you do that, Dave 
Bastard Programmer from Hell 
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
