Hungarian UIs
-
Can't say I agree - changing the type of a variable is still very much a difficult job. The IDE will find and replace just fine for internal and private scoped variables, but for anything that's protected or public a change of type could break 3rd party code. This isn't really a problem solved with a good search and replace. Contrast this with swapping a type that maintains backwards compatibility, you'll give 3rd parties far fewer headaches.
You have to take what I said in context. This is about UI objects. I'd say the majority of developers working with UIs are not working on third party libraries. I also can't think of any good reason to directly use a third party library's UI variable, as anything of this nature is typically scoped (why would you have a public variable for a text box that your library is responsible for when you could keep write acces to yourself and provide read access via a getter in the off chance the party using your library needs it?) I suppose if you're using someone's library of custom UI views you may find issue here, sure, but that's not often done; such code is not likely to change the data type of the variables anyway as they are abstracted to suit the needs of everyone (ex. a third party library that has a pull down to refresh list view for android is not likely to suddenly change the list view to a scroll view.) There are of course exceptions to all generalized statements like the ones I made, but honestly, in context, do you see renaming a UI variable as being an issue? UI objects are typically obfuscated in libraries and within your own code, you can easily rename the variable.
-
You have to take what I said in context. This is about UI objects. I'd say the majority of developers working with UIs are not working on third party libraries. I also can't think of any good reason to directly use a third party library's UI variable, as anything of this nature is typically scoped (why would you have a public variable for a text box that your library is responsible for when you could keep write acces to yourself and provide read access via a getter in the off chance the party using your library needs it?) I suppose if you're using someone's library of custom UI views you may find issue here, sure, but that's not often done; such code is not likely to change the data type of the variables anyway as they are abstracted to suit the needs of everyone (ex. a third party library that has a pull down to refresh list view for android is not likely to suddenly change the list view to a scroll view.) There are of course exceptions to all generalized statements like the ones I made, but honestly, in context, do you see renaming a UI variable as being an issue? UI objects are typically obfuscated in libraries and within your own code, you can easily rename the variable.
99% of the time yes, but I have had to deal with projects which did reference internal UI components, and it was a nightmare to refactor when things needed changing. Admittedly this is an edge case and I agree with you for the most part, but I'm of the opinion that you should be as defensive about these things as possible, even when you don't anticipate problems like this.
-
I think we can all agree that for the most part nobody uses Hungarian notation for variables any more... but it still seems prevalent in UI programming - for example I might have lblSomething next to txtSomething. On one hand I feel a bit uneasy that there must be some way to avoid this horrible practice, but on the other hand lblSomething is clearly meant to be a label which is next to txtSomething, and I need a way to differentiate between them without ending up with two controls with the same name. What say you? Disclaimer: I don't consider this to be a programming question, more a question of what styles people like to use.
Just like those few people not afraid of expressing their opinion (against popular dogma) and actually using it for good, I say Hungarian notation is not bad, but good. But just like everything, it is only good when you not overuse it. I'm using it in many cases with some exceptions (when the name actually conveys the type already or when it is a number and I don't really care how many bytes it is or if it signed, etc.). EDIT: Oh, yes, forgot to mention that I'm not using the 'pure Hungarian' prefixes, I just use the principle, making sure the prefixes are unambiguous. What I believe happened, is the Hungarian notation (similar to goto operator) are being witch-hunted, which is triggered by some incompetent and overconfident tongue (or hand/pen, rather) of some wanna-be computer science study book writer (and I'm implying a bunch of quite well-known authors here if you wondered). Well, after that the sheep herd just follows. Please, have your own opinion, people, it is refreshing. Just use the practice you think helps creating clear, maintainable code, regardless of what 'they' say. :)
-
I think we can all agree that for the most part nobody uses Hungarian notation for variables any more... but it still seems prevalent in UI programming - for example I might have lblSomething next to txtSomething. On one hand I feel a bit uneasy that there must be some way to avoid this horrible practice, but on the other hand lblSomething is clearly meant to be a label which is next to txtSomething, and I need a way to differentiate between them without ending up with two controls with the same name. What say you? Disclaimer: I don't consider this to be a programming question, more a question of what styles people like to use.
Don't use the system Hungarian. IDEs and compilers see types. If you declare variable like "int X;" in a statically typed language, X is an int. You won't make it float or double by mistake. On the other hand, the application Hungarian notation is still a good practice. In neural network programming, an input-to-hidden weight is double, and so is a hidden-to-output weight. Compilers cannot detect an input-to-hidden variable holding a hidden-to-output weight by mistake. Therefore, anything that has to do with an input-to-hidden weight should be marked with "ih", and anything, with an output-to-hidden, should be marked with "ho". A mistake like "ihWeights[i][j] += alpha * hoPrevWeights[i][j];" cannot be detected by your compiler, but reading just that line, you see your mistake. You should not be adding hoSomething to ihSomething.
-
Just like those few people not afraid of expressing their opinion (against popular dogma) and actually using it for good, I say Hungarian notation is not bad, but good. But just like everything, it is only good when you not overuse it. I'm using it in many cases with some exceptions (when the name actually conveys the type already or when it is a number and I don't really care how many bytes it is or if it signed, etc.). EDIT: Oh, yes, forgot to mention that I'm not using the 'pure Hungarian' prefixes, I just use the principle, making sure the prefixes are unambiguous. What I believe happened, is the Hungarian notation (similar to goto operator) are being witch-hunted, which is triggered by some incompetent and overconfident tongue (or hand/pen, rather) of some wanna-be computer science study book writer (and I'm implying a bunch of quite well-known authors here if you wondered). Well, after that the sheep herd just follows. Please, have your own opinion, people, it is refreshing. Just use the practice you think helps creating clear, maintainable code, regardless of what 'they' say. :)
The silliest part, of course, is the entire subject is subjective; there is no metric for wether Hungarian notation is good or bad. There is no performance difference in prefixing your variables. Each person's ideas of what's more maintainable is different. It's like the curly brace placement debate all over again. XD
-
I have been trying to get away from Hungarian for a loooong time, but I still use txtPassword and butOK and labPasswordPrompt. The original idea of HN was good - and I think it does help to distinguish UI controls with a prefix. If only to make it easier to find in intellisense! :laugh:
The universe is composed of electrons, neutrons, protons and......morons. (ThePhantomUpvoter)
Yeah, I particularly find it useful... specially when you have lots of fields to fill... Typing txt and seing them all filtered in a neat intellisense list just makes it much easier to manipulate and avoids forgetting one of them.
To alcohol! The cause of, and solution to, all of life's problems - Homer Simpson ---- Our heads are round so our thoughts can change direction - Francis Picabia
-
I think there's general agreement these days that it's not a good style, particularly if you want to change the underlying type of something without having to do a lot of renaming, aside from the aesthetics of it. However, everybody has seen the arguments for and against ad nauseum but I was curious as to what people do in the UI. One other exception that springs to mind is IInterface, which I don't have any problems with.
There are two types of Hungarian notation. One, the original, which describes purpose and the other which many use, data types. The former has as use because it helps you understand the code and how it is used. The latter, which many people use is a waste of time, unless you are using a typeless language like BCPL, vbscript or javascript.
-
There are two types of Hungarian notation. One, the original, which describes purpose and the other which many use, data types. The former has as use because it helps you understand the code and how it is used. The latter, which many people use is a waste of time, unless you are using a typeless language like BCPL, vbscript or javascript.
-
But if you are serious about UI programming (in WinForms, WebForms, and Mobile Apps) you want to imply a TextBox (WinForms).
Gus Gustafson
If you drag a control from teh Data Sources , Microsoft will call it firstNameTextBox -- who am I to argue if you have camel humps in R# who really cares , verbose maybe but self explanatory
Mike
