Well as far as i understand, LN channels can be somehow shut down, via certain glitches or other, and in that case, what would remain of the operation made in that channel ?
And what validity would this journalizing have regarding on chain state if there are difference at the end ?
The problem with caching is not about HD crash, but if the controller is stopped before the cache is a actually wrote, even if the hard drive works well the data is still lost.
Just to push analogy with cache to show certain caveeat, with smp system and cpu cache, there are certain case when the memory is shared with other chips with dma, or virtual pagination, in system with high concurency on the data, cpu cache can become "out of date", even with sse2 there are certain thing to help dealing with this, but as far as i know, most os disable caching on certain shared memory because of all the issues with cache, and instruction reordering etc When having access to up to date data in concurent system is more important than fast access to potentially out of date data.
If LN is to be seen as a cache system , it doesn't look like they are taking all the precautions for it to be really safe.
Cache are easily safe when all the write access to the data are made throught the same interface doing the caching, which is not the case with bitcoin & LN.
With hard drive it works because all the access goes throught the same controller doing the caching.
But anyway as LN locks the bitcoin on the main chain, it's not even really a true cache system, because the principle of a cache system is to fasten multiple access on the same data, as the bitcoin are locked, the channel have exclusive access to it, and so it's not really to be seen as a true system of blockchain caching.
And what validity would this journalizing have regarding on chain state if there are difference at the end ?
The problem with caching is not about HD crash, but if the controller is stopped before the cache is a actually wrote, even if the hard drive works well the data is still lost.
Just to push analogy with cache to show certain caveeat, with smp system and cpu cache, there are certain case when the memory is shared with other chips with dma, or virtual pagination, in system with high concurency on the data, cpu cache can become "out of date", even with sse2 there are certain thing to help dealing with this, but as far as i know, most os disable caching on certain shared memory because of all the issues with cache, and instruction reordering etc When having access to up to date data in concurent system is more important than fast access to potentially out of date data.
If LN is to be seen as a cache system , it doesn't look like they are taking all the precautions for it to be really safe.
Cache are easily safe when all the write access to the data are made throught the same interface doing the caching, which is not the case with bitcoin & LN.
With hard drive it works because all the access goes throught the same controller doing the caching.
But anyway as LN locks the bitcoin on the main chain, it's not even really a true cache system, because the principle of a cache system is to fasten multiple access on the same data, as the bitcoin are locked, the channel have exclusive access to it, and so it's not really to be seen as a true system of blockchain caching.
There are multiple implementations of the payment channel schema. Of course they have different trade-offs.
You remind me of young GMAX proving Bitcoin was impossible, and I hope you are similarly happy when shown to be wrong about LN.
I thought you might be interested in this tweet; to me it seems there is an interesting congruence afoot. Convergent morphology perhaps....or simply Data Structures 101?
To paraphrase:
"What sucks about directly buying Frappuchinos with Bitcoin?"
"The biggest issue I think is random small tx are literally the work that kills Blockchain the most."
The message seems to be that without write caches we don't get to have nice things!
In terms of mechanical engineering, write caches function as a shim which reduces friction and the resulting heat/damage.
https://en.wikipedia.org/wiki/Shim_(spacer)