A fair bit of this also falls out of the fact that DNSMasq is already running, configured, and modularised much as I describe in the original post. The larger part of that describes how to get to the point of dropping in a single script-generated file to provide Sci-Hub access on literally any TLD on the Internet in a fraction of a second, for all managed systems.
Note that not all systems can use hosts files: Android devices and iOS are highly resistant to this. They would have to be individually configured. Using DNSMasq, any device acquiring network leases from the local DHCP server is automagickally configured. "It just works."
The idea occurred to me a few days ago, and I ginned up and tested the solution in about five minutes. It took ten times longer to actually document it.
I don't know. For addressing individual content via a hash, yes, but I've not been impressed by the underlying usability.
My view of IPFS is that it's an address/directory-level piece of infrastructure that requires a UI/UX layer on top of it that abstracts out the hashes, which are not particularly human-sensible.
The simpler solution here is to just put it in your HOSTS file. And you don't really need to put all of them in there; just put in sci-hub.org and use that.
It's the most common way of doing ad blocking on Android, you do it the same way as you would on a Linux PC. Of course, you need root, just like on a PC. There are apps that will do it for you even.
Is there any reason for having 1542 TLDs that point to the same address instead of just one, other than wasting time trying to conform to invented but pointless requirements?
You install Telegram, you contact the bot https://t.me/scihubbot, you send him a DOI and it will answer with the PDF (if it has it).
Sometimes it says (in Russian) there is an error fetching the paper: I understand that in such a case it does not have the paper, but the paper is available on GenLib, so you can go there. But it is not frequent.
That sounds as if it's adding another layer (and possible failure mode) between you and Sci-Hub. How does the bot itself access the content? What keeps it from being attacked as Sci-Hub has, by publishers?
SciHub is unfortunately rather opaque on its internals (and that is definitely something that I would like to see changed), but it seems that the Telegram bot is blessed by the official Twitter account. In practice, I have often found it more reliable than the website, and it does not have problems with finding the right TLD. Also, as I said, it retains the history and it shared is with other devices, provided that you consider that a plus.
SciHub has operated at a set of IPs for most or all of its existence, certainly the past several years so far as I'm aware.
So long as you resolve it by hostname at those IPs, by whatever means, you'll be able to reach it there. If the IPs themselves are accessible.
If Telegram works for you, great. Meantime, if some school or research organisation that cannot afford to pay Western prices for academic content cares to set up their own networking infrastructure to provide access similar to the methods I've suggested, also great.
31 comments
[ 3.5 ms ] story [ 78.7 ms ] threadSci-hub currently does not fully function via IP address.
I've just expanded on why hosts files or other options don't offer the power of DNSMasq:
https://www.reddit.com/r/Scholar/comments/7m3uin/meta_if_you...
A fair bit of this also falls out of the fact that DNSMasq is already running, configured, and modularised much as I describe in the original post. The larger part of that describes how to get to the point of dropping in a single script-generated file to provide Sci-Hub access on literally any TLD on the Internet in a fraction of a second, for all managed systems.
Note that not all systems can use hosts files: Android devices and iOS are highly resistant to this. They would have to be individually configured. Using DNSMasq, any device acquiring network leases from the local DHCP server is automagickally configured. "It just works."
The idea occurred to me a few days ago, and I ginned up and tested the solution in about five minutes. It took ten times longer to actually document it.
My view of IPFS is that it's an address/directory-level piece of infrastructure that requires a UI/UX layer on top of it that abstracts out the hashes, which are not particularly human-sensible.
Will your hosts file handle 1542 TLDs?
Does your device have a configurable hosts file?
Almost certainly a similarly configured iOS device.
(I use the former heavily. It is among the devices autoconfigured by the system described in the OP.)
https://www.youtube.com/watch?v=ITs-YX1yQ7o
Sometimes it says (in Russian) there is an error fetching the paper: I understand that in such a case it does not have the paper, but the paper is available on GenLib, so you can go there. But it is not frequent.
So long as you resolve it by hostname at those IPs, by whatever means, you'll be able to reach it there. If the IPs themselves are accessible.
If Telegram works for you, great. Meantime, if some school or research organisation that cannot afford to pay Western prices for academic content cares to set up their own networking infrastructure to provide access similar to the methods I've suggested, also great.