Finding Posts Using Tags in Steem is Broken - Steemit Inc. Dev Team Says They Won't Fix it?
Have you experienced difficulty in finding posts on Steem based websites when using tags? I have and I just found out it's a known bug.. But for some reason Steemit Inc. decided not to fix it..
This bug report was raised by @roadscape at the start of 2017 and highlights that some posts simply don't get displayed when trying to view the 'created' list for posts with a specific custom tag.
It was identified that the RPC call to 'rpc.get_discussions_by_created' returns the list with missing posts, so this is a bug with Steem and not Steemit or other sites. The 'created' path in steemit.com is intended to show all posts for relevant contexts, just like 'hot' shows newer, popular posts and 'trending' shows the most popular ones over a longer time period.
I was intending to go back through 2 years worth of posts on Steem to organise them in a much better way, now that we can edit posts older than 7 days (this feature is currently only available on testing sites but looks to work ok). However, after doing that for my first custom tag, I found that some of the posts are missing.
Have you ever seen this?
Do you know why @vandeberg just said "won't fix" on this totally valid bug!? It seems he hasn't posted to the blockchain at all in several months.. Does he still work for Steemit inc?
Anyone? @andrarchy? @ned?
Thanks.
Wishing you well,
Ura Soul
Vote @ura-soul for Steem Witness!
(Witnesses are the computer servers that run the Steem Blockchain.
Without witnesses there is no Steem, Steemit, DTube, Utopian or
Busy... You can really help Steem by making your 30 witness votes count!)
I discussed this.... a looooong time ago.
The biggest thing steem is missing is a proper search function.
You can't even look up your old posts if you have too many posts.
A long time ago, they were going to introduce a separate tab for resteems, however they found, after much programming, that the back end couldn't handle it. So, it was never implemented.
Similarly, you should be able to look up @ura-soul and #steem
and get a list of posts. You can't.
In fact, you are forced into trending when looking up a particular tag.
At this point, i am unsure if the blockchain handling witness computers can even do this kind of sorting. I am pretty sure that the way the blocks are made makes every attempt at sorting and searching to be the hardest possible path.
"At this point, i am unsure if the blockchain handling witness computers can even do this kind of sorting. I am pretty sure that the way the blocks are made makes every attempt at sorting and searching to be the hardest possible path."
One possible solution is to transfer this to a 2nd layer, i. e. don't rely on the raw (or base) blockchain itself to have the search function, but to build it as a 2nd complete layer through a "dapp" running or interacting on top of the base layer.
If I understand things correctly that is what "hivemind" is supposed to be. A layer that handles tags, follows, and other things that don't need to be immutably added to the chain. That might also be why the marked it as "wont fix" because they aren't going to fix it within the chain but as a part of a second project.
The last part is conjecture, but if you've dealt with many programs you know some lack social graces and this sounds like one of those issues.
That is likely, yes. It would be good if they would have explained that in the Github ticket though.. And we are still left guessing and unclear what to do with our own development in relations to such an important issue.
It sounds very plausible, from a programmer's perspective.
And yes, sorting and searching would very much need to be a layer on top of the blockchain.
OR, you just search through the entire chain for everything. (truncated of course... you hope)
So, basically building a sorted database as people access information. Is what i picture when i read "hivemind".
... i hope they are not intending to build a permanent database... the resources would get huge.
(and the only way this is not conjecture is to be sitting in Stinc and completely knowing the back end)
@Steemsql and other databases already exist as a database form of the blockchain and I use it for Steemocean.com - but yes, the resource load is high, so I hear.
This makes way too much sense... don't bother telling it to the guys at Stinc.
[sarcasm]
The 'created' paths already do display the list of correct posts, it's just that some are missing for a reason I don't yet understand. The issue with being forced into trending has more to do with the underdevelopment of Steemit.com - since it is entirely possible to default to view 'created' and to also allow for switching between the various lists, such as 'hot' and 'trending'.
I would consider developing code for this, but I hate making code that others have already made or are making. The lack of clarity regarding a features roadmap makes deciding to code for Steem difficult and is why I have deliberately stuck to creating novel features at SteemOcean.com that aren't likely to be made elsewhere.
I don't know about the processing overheads for performing such searches at volume. Perhaps they are waiting for the modularisation of the witness nodes (to allow for clustering) before proceeding with that.
The reason i haven't developed code for this is that i have never been acknowledged by any programmers at steem inc. So, since this would be code that would change a great deal the load and processing done by the witnesses,... i would never even start this journey without the go ahead from someone with the authority to implement it.
I am pretty sure the algorithm just starts at the last block, and then goes backwards through the blocks until it gets 20 and then returns the list. Which means anything that is not current runs through a lot of blocks.
There is that, too! But to be fair, there are many thousands of users and they have no real way to communicate with all of them or to know who they are.. I think the steemdev discord rooms might be the best way in to communicate - or github.. However, I rarely get replies from anyone at steemit on anything anyway.
There are 22 results combing back in the 'created' list for the tag I created for my witness posts - so I'm not so sure there's a numerical limit:
https://steempeak.com/created/urasoul-witness
You almost sound surprised.
If steem inc don't give a toss about steemit as a blogging platform, which is a well established fact, why would they bother fixing anything that limits its functionality as just that?!
Posted using Partiko Android
There is a difference between Steemit and Steem. They may not be developing steemit.com but that is justified by other teams developing their own UIs and allowing Steemit inc. to focus on Steem itself. Fixing Steem itself seems fundamentally necessary, regardless of what happens with steemit.com.
Do tags matter beyond steemit then? Maybe they do...
Posted using Partiko Android
Yes, all posts on the Steem blockchain benefit from tagging - so all websites that use Steem will tend to make use of tags too. I am currently testing out different ways of getting my posts directly from the blockchain using a different call to get the data. :)
You'd imagine that organizing content is the next step after generating and curating. I've been telling myself that there is value on Steemit in the dataset, and it is just a matter of sorting the gems from the other.
I'm starting to wonder if this is true.
I just discovered that DTube deletes videos after 2 months. I've been paid well for my videos, but it is hard for me to say that these posts are valuable because there is no longer content where there once was. It seems that I'm the only person it bothers that people (myself included) are being paid to generate basically nothing.
Posted using Partiko Android
DTube's deletion of videos is a major hurdle. Youtube apparently isn't even profitable due to the massive costs involved.
In terms of the text content of the Steem blockchain though, there is huge potential waiting to be manifested by way of the development of useful tools.
Thank you @ura-soul for bringing this issue forward again. I also have the impression that the search tool provided by Steemit.inc is not really properly coded. If I'm just looking for a simple post, by entering the title of the post, mostly I'm not getting the right results.
I guess like @ned stated they are right now only concerned with the Steem-Blockchain-Code and not with the frontend so I feel that this issue will remain for a longer time.
Anyways, good to hear that you have mentioned that problem!!!
This issue really crosses the border between Steemit and Steem, since the data is being returned from functions operated from the RPC servers that interface with Steem and it is not technically a steemit.com bug.
I recall reading on how they are working to make searching using 3rd party tools more effective, we shall see.
STEEMIT has a lot of fixing up to do and a lot of features to make @ura-soul just to make this platform more friendlier to its users.