The comparison is a little pears to apple. Similar nutritions but different enough to not draw conclusions. The hardware in the Ceph test is only capable of max 1.7TiB/s traffic (optimally without any overhead whatsoever).
I also assume that the batch size (block size) is different enough that this alone would make a big difference.
That difference is still pronounced, yes. But the workload is so different. Training AI is hardly random read. Still not a comparison which should lead you to any conclusions.
Archive formats are hard to make reproducible because there are lots of ways of making different yet equivalent archives.
So it’s not surprising to me that someone would fail at this hurdle and find it frustrating to resolve.
Nix defined their own format for this to avoid this exact problem.
It seems there are multiple reasons. For one, the apk files include a digital signature and you won't have Signal's and Google's private keys available to recreate their signatures.
Thank you for this nice response. Did you already know or did you look it up? please don't tell me you just copied and pasted my question into an input form somewhere and it gave a bunch of reasons...
Ah nice; they got rid of that explicit warning - instead though we have the entire section about "bundlePlayProdRelease" including an externally sourced binary blob.
I don't understand how the details of the build process matter if the resulting files can be checked to be bit by bit identical? I can only think of something like Signal and Google conspiring to backdoor the binaries during the build process via this external binary blob. But if Google is part of this, they could also do it within Android which is not fully open source.
If you don't like this, you use the non-Play Store build instead (which supposedly doesn't include any binary blobs, but I haven't checked).
It's not only potentially infested with vulnerabilities. It's also not possible to filter io_uring using seccomp at all. So if you allow io_uring, you allow all that is possible with it.
I bought a 1st generation framework in a pretty early batch. In between I upgraded to a ryzen mainboard and use the old mainboard as a desktop computer in the dedicated case.
The old battery is not as good as it were (even though I only charge it until 75%, which the bios lets me configure) and the display hinges are a little too weak. The hinges got fixed for newer models, even in the factory second 1st gen Intel I bought for a friend they are fixed.
There was a time when they could not keep up with bios updates, but that has apparently been sorted out. And it's not worse than other vendors...
IMO the big problem is that it's very hard (and AFAIK no way has even be proposed) to restrict from the outside which syscall-alikes are allowed via io_uring. This is trivial with regular syscalls. That's important if you ever co-host applications with different security domains.
Is there some documentation on how to integrate it into a different web app? I would really like to open a template that is stored in the database and handle sending the mail, "exporting" (rename it to "save"), etc.
You can integrate your api call in the exportHtml() method to save it to your database. And you can load the same template into the component by providing it to [template].
reply