Loading…
Newly tagged old posts not appearing #13
3 participants
When I type "Gopalkrishan" into the search box on spnetwork, I get taken to this page:
https://selectedpapers.net/arxiv?searchString=gopalkrishnan
which says:
[1]: A coercion-resistant protocol for conducting elections by telephone Manoj Gopalkrishnan [2]: On the Mathematics of the Law of Mass Action Leonard Adleman, Manoj Gopalkrishnan, Ming-Deh Huang, Pablo Moisset, Dustin Reishus [3]: Catalysis in Reaction Networks Manoj Gopalkrishnan [4]: A geometric approach to the Global Attractor Conjecture Manoj Gopalkrishnan, Ezra Miller, Anne Shiu [5]: A Projection Argument for Differential Inclusions, with Applications to Persistence of Mass-Action Kinetics Manoj Gopalkrishnan, Ezra Miller, Anne Shiu
These include the 2 posts by Gopalkrishan that he said were not picked up by the spnetwork. So, I've asked him where, exactly, he's not seeing them.
[EDIT: Never mind. Gopalkrishnan's 2 posts were about arXiv articles written by himself. Above, the spnetwork is picking up the existence of those articles... not his posts about them!]
The current polling mechanism only goes back to posts less than 10 days old. This is because Google+ gives no mechanism for searching for what's changed. hence to discover old posts that are newly tagged, the polling mechanism might have to plow through the entire history of ALL spnetwork posts. One simple workaround might be to add a button that lets the individual user make it scan his/her posts for newly tagged (old) posts. Would that work for you?
Yes, that would be great. Not only Gopalkrishnan but another person has already reported this issue, so we can expect a rash of people trying to test out spnetwork by adding hash tags to old posts they wrote. They will be happier if this is possible.
Grr, github seems to have eaten my last message where I described in detail our interface choices for this. I'm not going to retype all of that. Bottom line: when a user views their own "user profile" page on selectedpapers.net, add an option so the existing "Check Google+ for Updates" button will scan ALL their posts (no matter how old; currently it only scans posts up to 20 days old).
That sounds great! When this is done I'll advertise this feature.
it's done, and I tested it on Manoj's posts (on a test platform, not on the production site). I guess I'll push out the update with this and a number of other fixes at midnight. The site will be unavailable for a few seconds when I restart the web server process.
the fix is deployed to selectedpapers.net. Please check it, comment on whether we should close this issue.
Chris, the pulldown menu I see only shows "last 20 days", with no other options. I tried manually changing "recent" to "all" in the URL, and sure enough, four of my older posts have now appeared. However, there are still two posts that don't show up. Both are reshares. If I search for #spnetwork on Google+ and narrow the search to "From you", it shows all six. However, if I search for "#spnetwork Dan Christensen", only four of them show up. The two missing ones are:
https://plus.google.com/115742208154874814350/posts/Euy9YSqLgoy
https://plus.google.com/115742208154874814350/posts/DrKrBDGRCKe
Ok, the pulldown menu now shows "All", but those two posts are still missing.
A couple notes:
the interface only shows the "all" option for a user to check for updates from themselves. My rationale was that users who are impatient for updates from someone else presumably want to see that person's latest posts.
hmm, Google+ categorizes "reshares" as different from "posts". Furthermore, in my original development work I had to carefully eliminate reshares that were just duplicates of the original posts. Otherwise you end up with a stream of messages that have the same text. I obviously did that too aggressively -- right it just filters out all reshares. That's why those two reshares don't get indexed.
The question is how hard it'll be to tell the difference between reshares that are just copies of the original message text vs. reshares with novel content.
What I'd like to do is treat that as a new, separate issue. First of all because it affects not just this issue of "newly retagged old posts", but our whole indexing process (e.g. brand new reshares are also ignored). Second because there are deeper questions about how this should be handled, which we'll need to discuss and decide before we can even think of implementing it. I will create a new issue.
Manoj Gopalkrishnan wrote:
I asked if he had joined the spnetwork, and he says he has.