Disqus - Latest Comments for soniccat

Re: RxJava vs. Coroutines + Flow

soniccat — Thu, 04 Feb 2021 14:33:45 -0000

Hi, thanks for such a precise article.

I'm curious about using Coroutines and RxJava at the same time while switching from one to another in a project. I suppose there might be problems with having different schedulers not knowing of each other. For example, if we use only Coroutines or only RxJava a corresponding computation scheduler is aware about how many cores a device has and takes that into account while managing an operation queue. On the other hand, if we have the RxJava Computation scheduler and the Coroutines Dispatchers.Default, it seems together they might run twice as many operations than required. Hence, I expect that might lead to overall performance degradation for simultaneously run operations. Probably there's a similar problem with the RxJava IO scheduler and the Coroutines Dispatchers.IO but I'm not sure.

What do you think?

Re: MVI - The Good, the Bad, and the Ugly

soniccat — Sun, 18 Aug 2019 03:38:14 -0000

I'd add that for relatively simple but popular app screens where we show a list of data loaded from a server, the current Google approach based on using LiveData with Resource template type looks like passing ViewState to the view and rendering it. The state is updated in the ViewModel via methods (Actions). It seems MVI won't be useful in this scenario.

For more complicated examples that might not be true.

Ref to the Resource type: https://developer.android.c....

Re: The Hobby · Moments of Inertia

soniccat — Thu, 23 Nov 2017 04:16:53 -0000

Some of my painted models in 2005 year. It was a try to make a Russian Warhammer. It's called "Ring of Rules". But I suppose, because of poor rules the project was closed.

https://uploads.disquscdn.c... https://uploads.disquscdn.c... https://uploads.disquscdn.c...

Re: Swift: When to use guard vs if

soniccat — Sat, 25 Mar 2017 13:51:19 -0000

Prefer not to use guard at all because, first of all, it forces to write more than one return statement. It's easier to understand small functions with only one. Then guard forces to handle wrong cases with early exits which might be wrong and should be done in another place.

In above examples, I think it would be better to move presented case blocks to separate functions. Where each function has one if statement and one return statement.

Re: Why NULL is Bad?

soniccat — Mon, 19 Sep 2016 13:01:39 -0000

What do you think about using annotations like Nullable and NonNullable? It should help to avoid long names like getByNameOrNullIfNotFound. Just name it getByName and mark it Nullable. IDE should help with highlighting situations where you forget to handle null cases.

Also it's strange that you're against returning null from get methods for storages like a map. Because returning an empty result from an empty map or a database is a valid result. It's logical to return a NonNullable result for methods like createAccount and throw an exception if it isn't possible. But it's expected to have an empty result from a storage object.

If I assume that throwing an exception from such get methods is a good idea that will lead to strange code. For example: I wan't to show one kind of UI if a person exists in a storage and another kind of UI if he is not. According your advice I should put one case in an exception handling part which smells like go to...

Re: Handling Android resources with non-standard formats

soniccat — Wed, 07 Sep 2016 10:21:08 -0000

In my case only
<dimen name="match_parent">-1px</dimen>
<dimen name="wrap_content">-2px</dimen>

works.

Your solution with item gives Resources$NotFoundException.

Re: Custom Jackson Polymorphic Deserialization without Type Metadata - The Boy Wonders

soniccat — Sun, 27 Mar 2016 15:23:32 -0000

Thanks, that treeToValue method is what I was looking for.

Re: Converting Complex Objective-C Macros to Swift Functions

soniccat — Fri, 30 Jan 2015 16:24:35 -0000

It seems there's no way to convert string concatenation (##) to generate new class name for example to swift.

Re: KYKY.ORG : Бестактность наших дней

soniccat — Mon, 05 Jan 2015 02:14:30 -0000

Имхо если вопрос кажется дурацким то и отвечать на него надо по дурацки: типа "завтра", "вчера". А если все нормально то и отвечать как есть: "без понятия", "не знаю", "в следующем году". Не вижу проблемы.

Re: Calling Blocks Inline

soniccat — Fri, 28 Feb 2014 12:52:56 -0000

>it would have to be defined outside of the method and have all the used local variables as arguments

Yes, I don't see any problem there. Two small methods are better than big one.

>This approach allows us to avoid such measures when the block is needed in only one place

Hmm, It's really hard to predict that block will be needed in only one place. Yes, today it's obvious, but what about next week?. Creating a separate method will be nice preparing to code reusing.

Re: Calling Blocks Inline

soniccat — Fri, 28 Feb 2014 11:47:41 -0000

It's better to create functions which just return const values. Will be much easier to read.

Re: Темы для 359 - Радио-Т Подкаст

soniccat — Fri, 27 Sep 2013 13:30:29 -0000

multi-path TCP в ios7: http://qz.com/126642/apples...

Re: Open Source: Library Allowing For Automatic Synchronization Between Portrait And Landscape Views

soniccat — Tue, 13 Sep 2011 14:25:45 -0000

Are you familiar with autoresizingMask? Or if your layout is very difficult in different orientation you may write single function like -(void)updateUIForOrientation:(UIInterfaceOrientatione)orient and call it in willAnimateRotationToInterfaceOrientation:duration:. Don't fight with iOS frameworks simply use it!

Re: Hey International Gamers, Go Download 'Mirror's Edge' for Free (Not in All Markets)

soniccat — Tue, 04 Jan 2011 01:00:10 -0000

In Russian it's free, but think gameplay will be boring.

Re: Regular Expressions in Objective-C

soniccat — Tue, 03 Mar 2009 09:28:13 -0000

Excellent library, but when i use multi-byte character ( russian in my example ) function crash. That happen because pmatch[i].rm_eo store a number of byte in string. In my example russian character have 2 byte and in
self substringToIndex:pmatch[i].rm_eo

i have out of range exception because pmatch[i].rm_eo > [self length]
I write that for resolved that problem:

size_t buffSize = pmatch[i].rm_eo - pmatch[i].rm_so + 1;
char* buff = (char*)malloc( buffSize );
memset(buff,0,sizeof(char)*buffSize );
memcpy(buff,[self UTF8String]+pmatch[i].rm_so, pmatch[i].rm_eo - pmatch[i].rm_so );

[substring addObject: [NSString stringWithUTF8String:buff] ];
free( buff );