Mutex + Lock = CsLibGuarded

[@Bob-tu9jq]

Aside from your professional content, people are aware of your awesome English pronunciations. Your voices and tones are very good. 👍🏻

[@saulth]

Great, I made something just like this a few years back and it's been very nice to use, takes away many opportunities to fail. I like your implementations better, thanks 😁

[@jonkalb2746]

What a great library. Tell me again why it isn’t in Boost?

[@CopperSpice]

Glad to help you make this happen.

[@267praveen]

@@CopperSpice you didn't answer his question

[@CopperSpice]

@@267praveen, his comment was actually an inside joke. Jon offered to submit this to Boost a few years ago and has yet to make it happen. We are happy to support anyone who would like to pursue this endeavor.

[@KayFiiya]

Please share videos about how to use CopperSpice to build guis in c++

[@skarndev488]

Awesome simplification for the use of threading! Does CsLibGuarded allow sharing a singular internal mutex for multiple objects to avoid lock overhead?

[@CopperSpice]

I believe what you are asking is how to "guard" multiple pieces of data using the same mutex. The best approach to solve this is to place your data objects in a single structure. Then using the struct as the data type for the guarded variable. This is a really cool use case.

[@NoNameAtAll2]

Why doesn't such a respectable channel use universal initialisation at slide at 4:10?

[@CopperSpice]

As a correction, there is nothing in the standard called "universal initialization." This is a common misconception. There are several different forms of initialization which use braces, such as direct initialization, default initialization, zero initialization, and aggregate initialization. Some of these have interesting corner cases. We write our examples to highlight the current topic rather than teaching multiple things at once.

[@VishalSharma-ys9wt]

I felt like "data race" would've been the more appropriate term instead of "race condition" at some places in the video.

[@kormisha]

In lr_gurded class, updating write value does not need to block the readers but would not updating the read value have to block the readers?

[@CopperSpice]

In the lr_guarded class the two values are called "left" and "right". Can you tell me what you are referring to by the "write" and the "read" value?

[@kormisha]

@@CopperSpice as per the video, the write value is the one that gets updated with the mutex. The read value is the one consumed by the threads and the video says no mutex is needed as only read operations are performed. But I do not understand how you keep both values in sync without locking both at some time.

[@CopperSpice]

Thanks, now we understand your question. Let me give you an example of how the lr_guarded algorithm works. If the left value is currently being read then the right value can be updated by the writer. After the right value is updated, internal pointers are updated so new readers start reading the right value. After all old readers have completed, the left value is then updated to match the right value and this can be done without a lock. This writer may have to wait a long time for all old readers to finish. If you have other comments or questions please feel free to ask. This discussion added a lot to the video.

[@wizardy6267]

Another nice video

[@CopperSpice]

Glad you liked the video.

[@NoNameAtAll2]

This lib reminds me of rust's "mutex" type, but slightly better. Glad it came to c++

[@CopperSpice]

The initial release for CsLibGuarded was about 5 years ago. We have other videos which talk about the internals, this was our first "how to use" it.

[@manuelm.8364]

The Github link doesn't work. But the video is very interesting.

[@CopperSpice]

Here is the exact link to the source code for CsLibGuarded. github.com/copperspice/cs_libguarded

[@manuelm.8364]

@@CopperSpice Yes, I found it. I thought you may want to fix the link. There is a HTML 404 error.

[@CopperSpice]

@@manuelm.8364 Oh, your comment was referring to a bug in the youtube link. Seems they damaged the url we supplied. We will get this fixed.