Microsoft killing off the old Skype 7 for Windows .... for real this time

[Syntaxvgm]

2 hours ago, AluminiumTech said:

Why? Do you think developers should make native apps for 3 different platforms and spend more money in the process?

dumpster fires are cheap too, and if you keep the doors open you wont even melt the dumpster. 

[poochyena]

12 hours ago, Ryan_Vickers said:

Some would say that not being able to find them through search is a feature, not a problem.  You give out your info to those who need it rather than it being available to anyone interested.

 

Also I don't really know what you mean by custom usernames since you can change your username at any time, and you can set a nickname on a per server basis.

On skype, thats fine, you can just turn your username off of searches so they don't appear.

You can change your display name not username. Your username is given some bs random generated numbers at the end.

[vanished]

39 minutes ago, poochyena said:

On skype, thats fine, you can just turn your username off of searches so they don't appear.

You can change your display name not username. Your username is given some bs random generated numbers at the end.

Those numbers change if you change the text part

[LAwLz]

3 hours ago, ScratchCat said:

That's a rather bold statement, there are 3.3 million apps in the Google Play store of which only one needs to be well made to invalidate it.

I was being a bit hyperbolic, but you can barely call Android apps Java anymore. They don't function like Java does on let's say Windows (because of ART).

Plus if those apps were written in some other language then chances are they would be able to perform better, and use less power to accomplish the same tasks.

 

Hell, performance was one of the major reasons why Android NDK got made. It used to be pretty much impossible to get low-latency audio processing in Android without using OpenSL through the NDK. Google had done a lot of work in that area to get the latency for the Java code down, but OpenSL is still king as far as latency goes for audio.

Another example where Andorid NDK is far superior to the Java code is for graphics. If you use Java then you will be limited to the OpenGL ES APIs. If you use NDK you get access to Vulkan.

 

 

4 hours ago, ScratchCat said:

I'm not too sure how your source tested VS Code but I certainly cannot reproduce the ~20 second load time. As you can see below there is hardly a delay when opening the file and the memory usage for the 6MB XML file is around the same as the source obtained from opening a 60b C file. 

You can't really say "it didn't take 20 seconds on my computer" because you most likely don't have the same PC as him. I am not sure he had the program started either. Since he talks about startup time, my guess is that he had the programs completely shut down (probably not cached either, considering the long "startup time" results) and then opened a file in conjunction with starting the program.

 

As for memory usage, I don't know how to read that or it everything is showing. The test in the blog said Code used 349 MB of RAM when opening the 60 byte file and you said your test showed the same. The Code process I see in that video of yours uses 152MB, so I assume there are some other child processes using the other memory, bringing it up to the ~350MB?

 

In any case, that is a lot and would be far less if it used native code (like Sublime).

 

 

4 hours ago, ScratchCat said:

Yes Electron is not perfect, others have proven it can be improved and it would be better if Electron programs could share certain processes such as for the GPU the overhead could be reduced but if it frees up resources as it is easier to ensure cross platform compatibility to work on more important parts of an application it is worth it.

Saying that Electron is good because it works everywhere and is easy to develop for is like saying anal sex is the best because it works on all genders. Sorry but I don't like getting butt fucked because developers are lazy and want to take the easy way out.

 

I get that Electron has its place. It makes it very easy to make a cross platform program, and you can hire some front-end web developer to make your program. For a developer who is cobbling together some hobby project it's great. It's easy and fast to use.

But we're not talking about some little independent developer here. We're talking about Microsoft, and a program (Office) which generates over 24 BILLION dollars in revenue every year. Don't you think they could spend a little bit of that 24 billion dollars to make something a bit better and more high performing?

 

I mean, Sublime is developed by two random Australian dudes, and they have managed to make a cross-platform text editor which seems to be running circles around VS Code.

[Jtalk4456]

On 10/1/2018 at 8:17 AM, iLostMyXbox21 said:

they finally found Discord

and along with it they found all the hentai. 

[iLostMyXbox21]

1 minute ago, Jtalk4456 said:

and along with it they found all the hentai. 

lol that’s also true

[ScratchCat]

1 hour ago, LAwLz said:

Plus if those apps were written in some other language then chances are they would be able to perform better, and use less power to accomplish the same tasks.

ART on Android 6+ can match if not exceed C code in terms of raw performance and can almost certainly handle the vast variety of SoC configurations (Arch, DSP, NPU etc).

 

For graphics maybe other languages still have the advantage but the raw performance gap is almost neglegible now.

 

1 hour ago, LAwLz said:

You can't really say "it didn't take 20 seconds on my computer" because you most likely don't have the same PC as him. I am not sure he had the program started either. Since he talks about startup time, my guess is that he had the programs completely shut down (probably not cached either, considering the long "startup time" results) and then opened a file in conjunction with starting the program.

Maybe 1-2 seconds on a i5 450M which is weaker than the test machine, waiting for the highlighting to complete would be around 12s.

 

The linter made the difference, the article didn't mention one so I tested without but the test repository mentioned one:

Quote

Visual Studio Code jumps to the end of the file quickly, but then takes many seconds for the highlighting to complete.

 

1 hour ago, LAwLz said:

As for memory usage, I don't know how to read that or it everything is showing. The test in the blog said Code used 349 MB of RAM when opening the 60 byte file and you said your test showed the same. The Code process I see in that video of yours uses 152MB, so I assume there are some other child processes using the other memory, bringing it up to the ~350MB?

 

In any case, that is a lot and would be far less if it used native code (like Sublime). 

Yes the total was around 350MB. It seems the editor hardly consumes any more memory if more/larger files are opened.

 

While that may be true and it certainly would be nice to see an improvement I am going to wager that 99.9% of people using VS Code (excluding Raspberry Pi users who don't seem to have a package for Sublime or VS Code) will have more than 1GB of memory if they are actually developing on the machine to run a browser alongside the editor. Not an excuse for the usage but just a consideration.

 

[dfsdfgfkjsefoiqzemnd]

I have this one friend who is still on Skype because all his other friends still use it.  That's the only Skype contact I still have, and the only reason I still run Skype. 

 

Not sure I'm willing to tolerate Skype 8 on any of my rigs though. I gave it a try earlier this year and went back to Skype 7 because Skype 8 was complete and utter garbage. 

I'll probably delete my account by the end of this month.

[poochyena]

6 hours ago, Ryan_Vickers said:

Those numbers change if you change the text part 

if you pay for a subscription. Still can't get rid of it. I don't want numbers in my username.

[vanished]

1 hour ago, poochyena said:

if you pay for a subscription. Still can't get rid of it. I don't want numbers in my username.

Fair enough, but I think the vast majority couldn't care less about the numbers and regardless would gladly take it in exchange for all the other differences.  And as for them changing, they must have altered that then because it used to generate new numbers if you changed your name.

[poochyena]

36 minutes ago, Ryan_Vickers said:

Fair enough, but I think the vast majority couldn't care less about the numbers and regardless would gladly take it in exchange for all the other differences.  And as for them changing, they must have altered that then because it used to generate new numbers if you changed your name.

As I said, it is incredibly vital to my usability as I am unable to share my username with others irl since I can't remember the random numbers. I typically just tell people to search "poochyena" or "poochyenarulez" on what ever platform they want and they can talk to me there. Much easier that way. It doesn't work for discord though, so I never talk to people there.

[LAwLz]

8 hours ago, ScratchCat said:

ART on Android 6+ can match if not exceed C code in terms of raw performance and can almost certainly handle the vast variety of SoC configurations (Arch, DSP, NPU etc). 

I'd like to get back to what I said as soon as Android was brought up.

11 hours ago, LAwLz said:

I was being a bit hyperbolic, but you can barely call Android apps Java anymore. They don't function like Java does on let's say Windows (because of ART). 

 

Also, you really should read that article. Not only does is show C crushing Java in 2 out of the 3 tests, it is only in a single test, on a single version of Android, where Java is faster than C, and he has several explanations for it.

1) The program doesn't trigger the garbage collector. You will be very hard pressed to a real world example where this is the case.

2) The Java compiler might have found a shorter way of executing the code which the developer or the C compiler are missing.

 

At the end of the day, there are fundamental parts of Java (at least on Windows) which means that no a Java program can ever run as well as a properly optimized C program.

It will be much easier to write the Java program, and it will work on a variety of operating systems, but it won't be able to run as well. A garbage collector can not, and it will never be, as fast or efficient as properly optimized manual memory management in C.

 

9 hours ago, ScratchCat said:

For graphics maybe other languages still have the advantage but the raw performance gap is almost neglegible now.

What tests are you looking at? If Intel released a processor which had a 60% IPC increase then I wouldn't say that jump was "almost negligible". That's the difference in the SHA1 test you linked. As the size and execution time grows of the program, so will the negative impact of having to rely on a garbage collector.

 

 

9 hours ago, ScratchCat said:

Maybe 1-2 seconds on a i5 450M which is weaker than the test machine, waiting for the highlighting to complete would be around 12s.

 

The linter made the difference, the article didn't mention one so I tested without but the test repository mentioned one:

That explains the big difference. So in the end, Code is much slower.

 

9 hours ago, ScratchCat said:

Yes the total was around 350MB. It seems the editor hardly consumes any more memory if more/larger files are opened.

First of all, 350MB is a lot for that program. Again, it's bloated as hell. Compare it to Sublime (which is very on par with Code in terms of features) and you will see the big difference.

Secondly, the reason why "the editor hardly consumes any more if larger files are opened" is because the program itself requires so much memory. If a program only needs 1MB of RAM and then it fetches and caches a file that's 1MB, that will increase memory consumption by 100%.

If a program uses 100MB and then fetches a 1MB file the consumption will go from 100 or 101MB, an increase of 1%.

 

Which program do you think is more efficient? The one that had a 100% increase in memory consumption, or the one which had a 1% increase in memory consumption?

 

9 hours ago, ScratchCat said:

While that may be true and it certainly would be nice to see an improvement I am going to wager that 99.9% of people using VS Code (excluding Raspberry Pi users who don't seem to have a package for Sublime or VS Code) will have more than 1GB of memory if they are actually developing on the machine to run a browser alongside the editor. Not an excuse for the usage but just a consideration.

I agree that it won't be a problem for most people, but that's my point.

This is the "good enough" mentality I was talking about before, where developers rely on better hardware to make up for their slow and bloated programs. The better hardware we have gotten, the worse optimized our programs has become.

[ScratchCat]

4 hours ago, LAwLz said:

I'd like to get back to what I said as soon as Android was brought up. 

I was referring to this point, it may be now that everything is an "app" there was some confusion but I meant the Android implementation of Java now can meet the performance of C in a number of cases.

Quote

Plus if those apps were written in some other language then chances are they would be able to perform better, and use less power to accomplish the same tasks.

 

4 hours ago, LAwLz said:

Also, you really should read that article. Not only does is show C crushing Java in 2 out of the 3 tests

C crushes Java

Negligible difference.

Minor difference.

I don't see the second "C crushes Java" test result (Not including the 32 bit versions as every phone made since circa 2013 has A53s and above).

4 hours ago, LAwLz said:

on a single version of Android

Only the latest version of Android matters and see above in regards to the 32 bit version.

4 hours ago, LAwLz said:

1) The program doesn't trigger the garbage collector. You will be very hard pressed to a real world example where this is the case.

2) The Java compiler might have found a shorter way of executing the code which the developer or the C compiler are missing.

1) True

2) That is the job of the compiler, to find the fastest way to execute the program.

 

5 hours ago, LAwLz said:

What tests are you looking at? If Intel released a processor which had a 60% IPC increase then I wouldn't say that jump was "almost negligible". That's the difference in the SHA1 test you linked. As the size and execution time grows of the program, so will the negative impact of having to rely on a garbage collector.

I took the average of the three shown tests but did not account for the lack of garbage collection in the second and third ones.