Unions in C#
-
I was old enough to touch PASCAL & COBOL. but FORTRAN -No, I missed it by a whisker. My previous batches did have this in their syllabus, in schools. :-O
Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
-
oops just seeing it's not there natively. (Not sure why) but done through interop :)
using System.Runtime.InteropServices;
[StructLayout(LayoutKind.Explicit)]
struct ByteArray {
[FieldOffset(0)]
public byte Byte1;
[FieldOffset(1)]Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
Damn socialists unions, they've taken over C# !!!
I'd rather be phishing!
-
oops just seeing it's not there natively. (Not sure why) but done through interop :)
using System.Runtime.InteropServices;
[StructLayout(LayoutKind.Explicit)]
struct ByteArray {
[FieldOffset(0)]
public byte Byte1;
[FieldOffset(1)]Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
Vunic wrote:
oops just seeing it's not there natively. (Not sure why)
Union seems to me more or less meaningless in managed code. At least it it contrary to the philosophy of managed code: You don't have a "memory address" but a handle. The runtime system will interpret that handle, translating it to "some" memory address that you shouldn't care about at all. The runtime system has the right to to that handle-to-address mapping in whatever way it pleases: It can do a JIT compilation (for code handles) and put the result into the assembly cache, it can load the data object from some backing store (in principle even magnetic tape!) etc. A union says: Let these two logical objects (usually fields in a class/struct), the two handles, be translated to the same physical address. That would strongly restrict the freedom of the mapping operation of the runtime system. What if the alternative interpretations have different sizes? What if one alternative contains a partial object, addressed through a handle, while another is a flat array? What if one member is an accessor function? Or a delegate? What about a list object unionized with a string? If the semantics of overlaying wasn't unambiguously defined, it would be useless, so you would have to define all the details of cases like these mentiones, and the runtime system would have to imlement exactly that (rather hairy) semantics. I am really happy that we have kicked unions out the back door. If you really need that functionality, then you declare a byte array, to hold the "memory block", and define accessors that select parts of that byte array, and by shifting and masking and explicit casting provides other interpretations of the "memory block". It is not done behind the curtain, where the reader would have to know how the compiler and runtime system works on the bit level. Rather, you code it all yourself, explicitly and with a big warnig sign: Danger! Arbitrary, unchecked casts made here! Then you define the overlaying in every detail, completely independent of how the compiler and runtime system treats your data. Yes, it will cost you a lot of work. It should. To keep you away from doing it.
-
Vunic wrote:
oops just seeing it's not there natively. (Not sure why)
Union seems to me more or less meaningless in managed code. At least it it contrary to the philosophy of managed code: You don't have a "memory address" but a handle. The runtime system will interpret that handle, translating it to "some" memory address that you shouldn't care about at all. The runtime system has the right to to that handle-to-address mapping in whatever way it pleases: It can do a JIT compilation (for code handles) and put the result into the assembly cache, it can load the data object from some backing store (in principle even magnetic tape!) etc. A union says: Let these two logical objects (usually fields in a class/struct), the two handles, be translated to the same physical address. That would strongly restrict the freedom of the mapping operation of the runtime system. What if the alternative interpretations have different sizes? What if one alternative contains a partial object, addressed through a handle, while another is a flat array? What if one member is an accessor function? Or a delegate? What about a list object unionized with a string? If the semantics of overlaying wasn't unambiguously defined, it would be useless, so you would have to define all the details of cases like these mentiones, and the runtime system would have to imlement exactly that (rather hairy) semantics. I am really happy that we have kicked unions out the back door. If you really need that functionality, then you declare a byte array, to hold the "memory block", and define accessors that select parts of that byte array, and by shifting and masking and explicit casting provides other interpretations of the "memory block". It is not done behind the curtain, where the reader would have to know how the compiler and runtime system works on the bit level. Rather, you code it all yourself, explicitly and with a big warnig sign: Danger! Arbitrary, unchecked casts made here! Then you define the overlaying in every detail, completely independent of how the compiler and runtime system treats your data. Yes, it will cost you a lot of work. It should. To keep you away from doing it.
-
*That useful reply* in lounge threads. :) :thumbsup: thanks
Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
Vunic wrote:
*That useful reply* in lounge threads
:confused: Such a thing exists? :laugh:
"the debugger doesn't tell me anything because this code compiles just fine" - random QA comment "Facebook is where you tell lies to your friends. Twitter is where you tell the truth to strangers." - chriselst "I don't drink any more... then again, I don't drink any less." - Mike Mullikins uncle
-
Vunic wrote:
*That useful reply* in lounge threads
:confused: Such a thing exists? :laugh:
"the debugger doesn't tell me anything because this code compiles just fine" - random QA comment "Facebook is where you tell lies to your friends. Twitter is where you tell the truth to strangers." - chriselst "I don't drink any more... then again, I don't drink any less." - Mike Mullikins uncle
-
oops just seeing it's not there natively. (Not sure why) but done through interop :)
using System.Runtime.InteropServices;
[StructLayout(LayoutKind.Explicit)]
struct ByteArray {
[FieldOffset(0)]
public byte Byte1;
[FieldOffset(1)]Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
var unioned = new[] { "Hi" }.Union(new[] { "I'm unioned" });
From the top of my head, but it's a union and it should compile :D
Best, Sander sanderrossel.com Continuous Integration, Delivery, and Deployment arrgh.js - Bringing LINQ to JavaScript Object-Oriented Programming in C# Succinctly
-
I thought Nobody Screws With The Union[^].
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
We have Union in the Netherlands :-\ Union is the city bike brand[^]
-
oops just seeing it's not there natively. (Not sure why) but done through interop :)
using System.Runtime.InteropServices;
[StructLayout(LayoutKind.Explicit)]
struct ByteArray {
[FieldOffset(0)]
public byte Byte1;
[FieldOffset(1)]Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
-
My first real programming job was all FORTRAN. 'Cause Real Scientists Program In FORTRAN. It was especially fun when everyone insisted that variables be named a,aa,aaa,b,bb,bbb,b1,bb1 etc. For a bunch of smart people they were truly awful at organising (let along thinking through) code.
cheers Chris Maunder
-
My first real programming job was all FORTRAN. 'Cause Real Scientists Program In FORTRAN. It was especially fun when everyone insisted that variables be named a,aa,aaa,b,bb,bbb,b1,bb1 etc. For a bunch of smart people they were truly awful at organising (let along thinking through) code.
cheers Chris Maunder
Chris Maunder wrote:
Cause Real Scientists Program In FORTRAN.
And you were using it why?
Michael Martin Australia "I controlled my laughter and simple said "No,I am very busy,so I can't write any code for you". The moment they heard this all the smiling face turned into a sad looking face and one of them farted. So I had to leave the place as soon as possible." - Mr.Prakash One Fine Saturday. 24/04/2004
-
My first real programming job was all FORTRAN. 'Cause Real Scientists Program In FORTRAN. It was especially fun when everyone insisted that variables be named a,aa,aaa,b,bb,bbb,b1,bb1 etc. For a bunch of smart people they were truly awful at organising (let along thinking through) code.
cheers Chris Maunder
My very first programming was in BASIC in those days when it was really BASIC: You had a maximum of 286 numeric variables named A to Z or A0 - A9 to Z0 - Z9, plus 26 string variables A$ to Z$. And then you could have 26 arrays, I believe they were named A# to Z#. I really should dig up that old "Real Time Basic for the Univac 1100 series" manual from my basement, for a nostalgia trip :-) That was in my high school days. At the Univerisity we learned Fortran and then Simula and Pascal, and the professor went to extremes in demanding long, descriptive variable names: When adding two numbers, fitting the names of the sum and the two addends on a single line could be a problem within the 80 char screen width. But the professor insisted. Then, I looked over the shoulder of the brightest guys in the class while he was typing in some Pascal code, and I gasped: You can't hand in that! ... Variables were named I01, I02... F01, F02 (F for Float) and so on. "Of course I won't!" he replied, "Before I hand it in, I have the editor automatically replace I01 with NumberOfApplesPerBasket, F02 with AverageWeightPerAppleInGrams and I02 with NumberOfBasketsPerLoad - I can't waste my time typing those terribly long names every time I use that variable!" -- Sure he was the brightest guy. To him, the mental effort of knowing the meaning of I01 was no greater than knowingt the meaning of NumberOfApplesPerBasket. So why not save a little typing? For my own part, I am happy with descriptive names (within limits), even if it takes a little more typing.
-
Chris Maunder wrote:
Cause Real Scientists Program In FORTRAN.
And you were using it why?
Michael Martin Australia "I controlled my laughter and simple said "No,I am very busy,so I can't write any code for you". The moment they heard this all the smiling face turned into a sad looking face and one of them farted. So I had to leave the place as soon as possible." - Mr.Prakash One Fine Saturday. 24/04/2004
You hurt me. You hurt me deep.
cheers Chris Maunder
-
My very first programming was in BASIC in those days when it was really BASIC: You had a maximum of 286 numeric variables named A to Z or A0 - A9 to Z0 - Z9, plus 26 string variables A$ to Z$. And then you could have 26 arrays, I believe they were named A# to Z#. I really should dig up that old "Real Time Basic for the Univac 1100 series" manual from my basement, for a nostalgia trip :-) That was in my high school days. At the Univerisity we learned Fortran and then Simula and Pascal, and the professor went to extremes in demanding long, descriptive variable names: When adding two numbers, fitting the names of the sum and the two addends on a single line could be a problem within the 80 char screen width. But the professor insisted. Then, I looked over the shoulder of the brightest guys in the class while he was typing in some Pascal code, and I gasped: You can't hand in that! ... Variables were named I01, I02... F01, F02 (F for Float) and so on. "Of course I won't!" he replied, "Before I hand it in, I have the editor automatically replace I01 with NumberOfApplesPerBasket, F02 with AverageWeightPerAppleInGrams and I02 with NumberOfBasketsPerLoad - I can't waste my time typing those terribly long names every time I use that variable!" -- Sure he was the brightest guy. To him, the mental effort of knowing the meaning of I01 was no greater than knowingt the meaning of NumberOfApplesPerBasket. So why not save a little typing? For my own part, I am happy with descriptive names (within limits), even if it takes a little more typing.
Classic! Why solve a problem when you can create a sub-problem and solve that on top of the main problem. Ah, software developers.
cheers Chris Maunder
-
oops just seeing it's not there natively. (Not sure why) but done through interop :)
using System.Runtime.InteropServices;
[StructLayout(LayoutKind.Explicit)]
struct ByteArray {
[FieldOffset(0)]
public byte Byte1;
[FieldOffset(1)]Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
I did like unions for things like: union ColorMashup { long ARGB; struct { byte alpha; byte red; byte green; byte blue; } } where the same data could be accessed as a whole or part. Polymorphism solves most "union" problems a lot better, but the memory will probably be spread around a lot more. You had to squeeze every drop of memory and performance out of the old unix machines.
-
oops just seeing it's not there natively. (Not sure why) but done through interop :)
using System.Runtime.InteropServices;
[StructLayout(LayoutKind.Explicit)]
struct ByteArray {
[FieldOffset(0)]
public byte Byte1;
[FieldOffset(1)]Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
Most useful for hardware registers and even software registers. You can view register broadly via the long/int/short or via the fields contained within it. Save a copy: myCopyReg.bits = mySrcReg.bits. As a bonus your get a copy of the "undefined" bits. Beats myRegCopy.fld0 = myRegSrc.fld0; myRegCopy.fld1 = myRegSrc.fld1; myRegCopy.fld2 = myRegSrc.fld2; et. nauseum
-
oops just seeing it's not there natively. (Not sure why) but done through interop :)
using System.Runtime.InteropServices;
[StructLayout(LayoutKind.Explicit)]
struct ByteArray {
[FieldOffset(0)]
public byte Byte1;
[FieldOffset(1)]Starting to think people post kid pics in their profiles because that was the last time they were cute - Jeremy Falcon.
Thank you for this posting. I am currently working on the conversion of some very old FORTRAN software for the NASA Voyager data processing as we are one of the instrument teams still working the Voyager Interstellar Mission and I have been struggling to find an option in C# to do exactly this, i.e. as mentioned in a reply this is like the FORTRAN COMMON BLOCK statement. Now I think I will be able to finish my conversion project a lot easier than attempting to write my own version of a COMMON statement and figuring out how to implement all of the various possibilities.
Jerry W. Manweiler, Ph.D. Fundamental Technologies, LLC
-
Thank you for this posting. I am currently working on the conversion of some very old FORTRAN software for the NASA Voyager data processing as we are one of the instrument teams still working the Voyager Interstellar Mission and I have been struggling to find an option in C# to do exactly this, i.e. as mentioned in a reply this is like the FORTRAN COMMON BLOCK statement. Now I think I will be able to finish my conversion project a lot easier than attempting to write my own version of a COMMON statement and figuring out how to implement all of the various possibilities.
Jerry W. Manweiler, Ph.D. Fundamental Technologies, LLC
-
var unioned = new[] { "Hi" }.Union(new[] { "I'm unioned" });
From the top of my head, but it's a union and it should compile :D
Best, Sander sanderrossel.com Continuous Integration, Delivery, and Deployment arrgh.js - Bringing LINQ to JavaScript Object-Oriented Programming in C# Succinctly
