kice

too many unexpected loose indentation warnings for marcos

1

I just don't know why I got loose indentation warnings here. I have some fairly complicated marcos and it generate way too many "loose indentation" warnings when compile even I...

``` // #name enum_test using System; public enum A { B, C, D, } // crash if type of a is enum public class Test { public A a; static...

希望增加下弹幕密度

4

弹幕密度不够啊。就是位置满了之后，继续加。 

两条一样的弹幕会崩溃

1

`qwe` `qwe` 这样就会崩溃。

There are also some other recipes can successfully add to the server. Those recipes I can add them when I play solo but cannot be added to server. ``` recipes.addShapeless(...

[PTRun] - only show top window per process in WndWalker

1

## Summary of the Pull Request Add a feature that allow to only show most top window (or "most recent window") for each process in Window Walker. ## PR Checklist...

我这边存在这样的情况: 最快的DNS查询结果不一定是最好的。 例如，垃圾ISP的DNS秒回，但是查询出来的结果太糟糕。或者同时查询1.1.1.1 和 8.8.8.8，google的返回结果更快不过解析的IP不行，但是CF的解析的IP更好。 查询DNS短时间内就一次，但是TCP数据传输会发生很多次，更好的解析结果比更快的响应更重要(大概)。 大体上设计思路: 1. `fast_forward` 增加 `waiting`，第一次返回结果后等待时间，单位毫秒 2. `fast_forward` 增加 `upstream - addr - weights`，权重，用于排序结果，无单位 在并发查询得到第一个结果后，等待 `waiting` 毫秒，然后检查等待期间是否有其他服务器返回解析。 然后根据全部查询得到的结果，按照返回的服务器权重进行排序，返回权重最高的查询结果。 复制一些的话，使用域名匹配器，可以根据域名来调整DNS权重。 其他替代方案 估计就是并发查询全部结果都缓存，然后后台TCP Ping检测查询结果是否高延迟或者低连通，并根据要求删除缓存。

Add C++ Predictor class for inference

19

## Description ## C++ Predictor class for easy inference * Support quantized model * Support non-float32 data input and output ## Comments ## Both c-predict-api and cpp-package are missing data...

We you select some words, it will highlight the same selected words in other lines.  I am not sure if the extension or just VS. But I am not...

It looks like it only supports function autocomplete and if it can also enable variable autocomplete, it will our lives much easier.