Do unread items sunset after 14 days?

  • 171
  • Question
  • Updated 5 years ago
I'm seeing unread items in NewsBlur change to read after 14 days of sitting in a feed, much as Google Reader would mark them read after 30. Is there a way to turn this behaviour off? I want persistent counts. NewsBlur already has a manual tool to unclutter feeds.
Photo of Elfpill

Elfpill

  • 28 Posts
  • 15 Reply Likes

Posted 7 years ago

  • 171
Photo of Juan Pablo Ortiz

Juan Pablo Ortiz

  • 9 Posts
  • 7 Reply Likes
There's any chance you can consider on disabling that intelligence feature optionally?

I don't really care for it, and I think there's more people like me who would like to enable permanent read statuses than use that feature.
Photo of Juan Pablo Ortiz

Juan Pablo Ortiz

  • 9 Posts
  • 7 Reply Likes
Can't agree more, Craig.
Photo of Josh Sled

Josh Sled

  • 123 Posts
  • 36 Reply Likes
Thanks for the correction.

Yes, speculative feature is speculative, and shouldn't block core functionality.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
I agree also, but this is Samuel's baby, and he has been pretty frank, honest, and open to ideas by the community, not just privately, but for the world to see. He is very open and transparent. We should probably consider respecting some of his vision for the service as long as he is compromising with us on the most important things to us, and also has maintained an otherwise slick and reliable service. I think if we can get at least a month of unread tracking that can get most of us by, though it would be much better to have substantially longer tracking, at least for feeds that are not very active.

I am personally just relieved he has agreed to give us a month as soon as he can. I will be grateful for that, though I respect everyone's opinion whether they agree or disagree with me.

Anyway, something to consider... Throwing Samuel a bone since he threw us a big one. (OK, yeah, that may not read well... :-)
Photo of Craig

Craig

  • 7 Posts
  • 5 Reply Likes
I don't begrudge him his vision of new tech ideas. After all, it could be that we just don't "get it" yet. In general, we're certainly impressed by what he has been able to put together.

My personal perspective, though, is that a solution or workaround to the quick expirations (becoming "read" when people haven't even had a chance to read them yet due to 2 week limit) should be a higher priority during the final weeks before Google Reader is shut down (even if maybe premium/paying subscribers are simply put on a different server with different settings).
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
Interesting idea, Craig, but I doubt that would be very practical / would introduce a lot of complications, like people upgrading/downgrading, fragmentation, etc. But I get your frustration. A majority of us probably have it. I certainly do.
Photo of Victor Orellana

Victor Orellana

  • 26 Posts
  • 2 Reply Likes
Remove the training feature once and for all
Photo of romkyns

romkyns

  • 44 Posts
  • 2 Reply Likes
I am personally still curious to see whether the impact on response time is measured and quantified, or simply suspected based on what the code does.

Are we suffering this because 30 days would double/triple the response time, or because it would add 10%?

Maybe you could sell 30 days upgrade to, say, 500 people and 90 days upgrade to 50 people. Surely you could handle that? And then crank up the number of allocations as the infrastructure grows. I'll pay! Heck, don't even code a UI for it; just ask us to set up a larger payment and set a flag in our accounts by hand.

In fact, I bet that the number of people currently willing to pay for this increase is low enough that you can: a) turn it on for everyone who wants to pay for it, b) keep the response times basically unchanged, and c) make some extra money in the process.
Photo of Craig

Craig

  • 7 Posts
  • 5 Reply Likes
+1 to what romkyns just said. Instead of worrying about budget constraints, make it pay for itself, and make those of us who care about it put their money where their mouth is to get what they want (just for themselves).
Photo of toupeira

toupeira

  • 36 Posts
  • 3 Reply Likes
+1 as well, unfortunately there's still the problem of this setting being set globally and not per user, and it seems due to the way the setting is used in other places the implementation will require a lot of refactoring.

I'd like to help with the code, but I can't find any developer guidelines anywhere and don't really want to bother setting up the huge amount of requirements to run a NewsBlur instance. Is there maybe a chance for contributors to test patches on dev.newsblur.com or something?
Photo of Splike

Splike

  • 136 Posts
  • 15 Reply Likes
Looking forward to them being kept unread for 30 days. Some people think that 30 days is not enough, but It's a whole lot better than 14, that's for sure. And for me 30 days is enough. I would never take a month vacation anyways.
Photo of Hikari

Hikari

  • 43 Posts
  • 18 Reply Likes
Guys, srsly. He doesn't want to, he doesn't care on what his paid customers need. Period.

He doesn't even care with competitors, he'd rather see us all move to Feedly, TinyRSS, or anything else. He also doesn't want the feature to profit.

Worse, with so many wanted features and fixes, he just ignore requests and does what he wants, which is making the software WORSE.

Face it. Newsblur is his personal RSS Reader and he lets us use it AS-IS. If we're happy with how he likes it, good. If we pay for it, great. If we want ANY change, just get out.

His happiness isn't on satisfying his customers. It's on refunding them and seeing they leave!
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
I agree this limitation should be disclosed more prominently before anyone buys (with the added comment that 1 month is coming as soon as possible), but at least Samuel refunds people who are not happy. I think it would make Samuel's life easier if he just disclosed it more prominently too. Not worth the hassle for him.

But, again, at least he lets you know where you stand in these comments instead of leading any of us on, and he is open to changing his mind about things based on customer feedback. For a one-man operation, I think this is about all we can ask for.

When developing a business like this, you just can't make everyone happy, and you have to make decisions. I think the frustration comes because there aren't many (if any) services that are otherwise as good, so it's not like there is a lot of good options.
Photo of happy.cerberus

happy.cerberus

  • 15 Posts
  • 4 Reply Likes
The fact, that he is offering refunds doesn't make this any less illegal. And yes, this kind of business practice is illegal in most EU countries (if you are wondering, yes, you must obey EU legislation if you are offering services in EU).

Also, the fact, that this has to be documented was brought up on numerous occasions in this thread, still completely ignored.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
I wasn't aware it was illegal if refunds are readily given out. But yeah, it is just the right thing to do... disclose it clearly before someone purchases, especially for such a big limitation of the service. As I said before, I would think prominent disclosure would be better and easier for Samuel too. It is just not worth all this for him,.

Of course, he does allow this forum thread to remain (and participates honestly himself), and does not delete reasonable comments, apparently, so the man is definitely not evil.

At this point, people should accept things as laid out by Samuel, or kindly ask for a refund and move on. I would think it not worth your time to hash this out any further if you cannot accept what he has said and the limitations he is working with (and his ideas about what he wants his service to be).

That said, I hope Samuel considers something clear about this limitation going forward for new people before they purchase.

Not sure I will respond to any more of these, and you guys are probably thrilled about that. It is getting to the point I am wasting my time and have nothing useful to add unless new information comes to light anyway.

Good exchange though. A pretty civil group, compared to most forums, though maybe the troll comments are being managed out so it only seems this way, I have no idea. Thanks everyone, and thank you Samuel for allowing public debate and dissent. You and your product will be better for it in the end.
Photo of Hikari

Hikari

  • 43 Posts
  • 18 Reply Likes
He isn't evil, he's not skilled to manage a business.

TinyRSS is also OpenSource and doesn't offer any kind of hosting. It has fewer features, but it has active dev community, and is simpler to install.

As I can see, because an architectural bad decision, it's simply too expensive to either increase unread limit or either fix the problem caused by that decision.

The software is unable to use different limits to different accounts. So, premium users willing longer limit are unable to pay the bill for the resource required to increase it to everybody.

I wish BitNami would build a setup for newsblur :/

Also, many ppl have complained they took months to find out why their articles are "disappearing". They probably came to GetSatisfaction to ask for support and found this thread. I guess how many are having this problem and haven't even noticed it yet, or noticed but didn't figure the cause.
Photo of romkyns

romkyns

  • 44 Posts
  • 2 Reply Likes
I don't share the view that Samuel is not skilled enough to manage this business. Also, I don't think you're helping the cause by annoying him. I understand you're annoyed (hey, so am I), but reciprocating is detrimental here...

I will point out that it also took me a while to figure out why stories were disappearing. Perhaps a couple of weeks. And I will once again ask for this behaviour to be documented. Politely :) Please, Samuel.
Photo of Malcolm Miles

Malcolm Miles

  • 1 Post
  • 0 Reply Likes
Just to add another call for no expiration of feeds. As others have said, I wouldn't have gone to a premium account if I had known of this "feature". Perhaps it should be documented more clearly (is it documented anywhere now)?
Photo of Benjamin Mako Hill

Benjamin Mako Hill

  • 7 Posts
  • 20 Reply Likes
I think things are getting a bit nasty and acrimonious in here. Thanks for keeping a level head Samuel. And also, thanks for offering to refund folks. I think that's a nice gesture. As somebody burned by this, but who really likes and supports the project, I appreciate your being responsive. We have different priorities, but I respect what you're trying to do.

And I know I've said this before but it's never been responded to: I really think you should document this in the FAQ or elsewhere. I'm able to work around this issue just by making sure I deal with my important feeds more regularly. But it took me months to realize why my unread articles were "disappearing" after I regularly checked the FAQ. I think if people knew about this limitation/feature, they would not be so surprised and would not feel so burned.

Thanks in advance.
Photo of pattulus

pattulus

  • 4 Posts
  • 0 Reply Likes
That's how to comment. Be nice, stay calm and be constructive.

= Mirrors exactly what I wanted to say.
Photo of Rob

Rob

  • 12 Posts
  • 0 Reply Likes
I concur - I probably wouldn't have signed up had I know about the 14 day expiration so it does feel a bit like I've been duped. The promise of a 30 day expiration is worth me hanging around for (for now) but it did feel a bit like bait-and-switch.
Photo of Victor Orellana

Victor Orellana

  • 26 Posts
  • 2 Reply Likes
Everyone, please be patient. Facebook will be rolling out RSS soon, and it will be awesome. I promise.

I have requested several times for help on problems I have encountered. Up to now, I have yet to receive a reply.
Photo of romkyns

romkyns

  • 44 Posts
  • 2 Reply Likes
Samuel, you're missing out on a lot of goodwill by not adding a mention of the 14 day limitation despite repeated requests. You *know* this confuses people. There's only so much time every day, but when people feel duped after paying you... perhaps it's worth prioritizing that. Besides, it's just a bit of text.
Photo of Victor Orellana

Victor Orellana

  • 26 Posts
  • 2 Reply Likes
Samuel, thanks for finally responding to me.

One of the issues right now is that after pressing shift-t, the stories column goes away, but comes back every time. Can you make it permanent?

Another issue is that I am using Firefox, and the "o" still opens pages in the foreground. I want them to open in the background.

Can you please look into these.

Thanks again.

Victor
Photo of Samuel Clay

Samuel Clay, Official Rep

  • 6514 Posts
  • 1474 Reply Likes
For Firefox, you'll have to change a browser setting: http://www.makeuseof.com/tag/links-op.... For the shift+t, I'm working on a preference.
Photo of Victor Orellana

Victor Orellana

  • 26 Posts
  • 2 Reply Likes
Thanks Samuel
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
As for Facebook, I hate that service, so I will have to be pretty desperate before going there, but you never know. They are pathological data and advertising whores. You cannot trust the privacy settings you make will be retained for long either. It is just a big time-sink waste of time the way most people use it. From what I can tell, the whole cult of Facebook is quickly wearing off. It is no longer considered very cool.

Some of their corporate giving (and policies) and personal giving by Zuck are to be commended though. I just hate the service itself in its current and past incarnations.
Photo of Craig

Craig

  • 7 Posts
  • 5 Reply Likes
-1 on: The belly-aching, so let's not continue it for now. We've already been told that if you want your soup to have 30 days and can't wait for that to be cooked-up by the master, it'll just have to be "no soup for you." Find a different product and make a different set of compromises.

+1 on: Yes, it does sound necessary to disclose technical limitations on the service that don't meet what people would expect. Sounds easy to do somewhere.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
PERFECT answer, Craig. Well done my man.
Photo of yardena

yardena

  • 35 Posts
  • 1 Reply Like
Samuel, my advice:

1) Fix the documentation bug NOW. That bug is critical and urgent. You have no good excuse not to. Failure to disclose poisons the water before any good faith debate can happen. Bug report comments don't count as documentation.

2) Write your responses carefully and mindfully, so older responses can be read at any time in any order, without causing misplaced frustration in the reader. Especially in this support interface which bubbles older responses to the top.
Photo of evancg

evancg

  • 11 Posts
  • 4 Reply Likes
I didn't realize this limitation existed when I signed up for premium a couple months ago. I was really confused as to why I seemed to be so current on my feeds - and then I started recognizing the "Oldest Unread" date was always about two weeks old.

Then I found this thread when I was looking through the Support and it all makes sense now. Bummed :( Wouldn't have shelled out $36 had I known. I like RSS readers because I don't have to worry about keeping current on every website I like to read, but with this limitation I'm basically forced to skim through at least every other week. Defeats the purpose...
Photo of Juan Pablo Ortiz

Juan Pablo Ortiz

  • 9 Posts
  • 7 Reply Likes
Anyone knows how many days is the unread limit on Feedly? or if it exists at all?
Photo of pattulus

pattulus

  • 4 Posts
  • 0 Reply Likes
Same as Feed Wrangler – they all seem to settle with 30 days.
Photo of laurie leonard

laurie leonard

  • 2 Posts
  • 2 Reply Likes
Some of my feeds on Feedly disappear after 14 days. Perhaps it depends on how active the feed is?
Photo of ianjo

ianjo

  • 7 Posts
  • 4 Reply Likes
I think the issue here is that there are (at least) two kinds of feeds that people follow.

One is the never-ending ever-growing one, like world news, that you glance at sometimes, but if you miss 3 weeks you're probably not going back to read more than a few days and give up. If you missed something on 9gag, don't worry, sooner or later it's going to be reposted.

The other is for following something that is usually much lower traffic, but you believe that is important you never miss an update. Maybe it's your favorite webcomic, where you read each entry and don't want to miss a part of a story. Maybe it's some friend's blog, or some tech blog detailing important info on something you use. You NEVER want to miss entries on these feeds.

This split is very clear to me: I used to keep the first kind of feeds in iGoogle, that only displayed the latest 10 entries (if I missed something, no problem), and the second kind I used rss2email to get it archived in my e-mail account so I could read whenever I wanted, if even a year late.

I hope the authors of Newsblur can recognize that this difference is very important, and either newsblur clearly states that doesn't care about the second case, and that it won't be handled, or (I hope so) it also covers the second use-case.

One of the discussed issues is that the Intelligence feature is one of the main reasons why this issue can happen. Again: Intelligence makes a lot of sense for the first kind of feeds, where they contain so much that you just want to filter a part; for the second part, I wouldn't mind not having the feature.

Finally, I haven't subscribed yet, but without this feature I can't see myself subscribing, as other people have said, this is *the* feature for a reader.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
Good analysis, ianjo. I agree with your main points.

However, Samuel has answered quite clearly that he does not intend to go past a month for unread tracking, largely due to technical constraints of NewsBlur's architecture (which he does not presently intend to change). If that is not enough for you, I guess you will have to move on. I think your feedback was useful though.

I would really like up to a year or more for unread tracking, but can live with a month. Most other services do not offer anything super long like a month either... Seems like about a month is pretty popular with other rss services, though I would think the average is 3 months or more, but that is what I get for thinking, I guess.

It might also help to remember that the two week or 1 month unread tracking on NewsBlur is just that... You can still see the items that age out, they just will look like they have been read (if my understanding is correct). So it is not the end of the world, but can be very inconvenient still. Thus this huge thread.

I am glad Samuel keeps us updated, but that he does not obsess about being too active here. I would rather he keep his head down and improve the service as fast as he can (in a sustainable way).
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
In my previous post, I see I made a mistake...
"Most other services do not offer anything super long like a month either"
S/B something more like
"Most other services do not offer anything significantly longer than a month either"
Photo of Hikari

Hikari

  • 43 Posts
  • 18 Reply Likes
Craig, for situations like yours, that have higher risk related to leaking some article, and/or are directly related to work/business, I'd strongly suggest a availability and reliability-oriented service. A public one for sure won't fit your availability and reliability requirements.

Try searching for an OpenSource software with lots of independent testers and dev contributors, they use to be the best in stability. Then get some good shared hosting service and setup in it. You'll be able to config the software as you want
Photo of Craig

Craig

  • 7 Posts
  • 5 Reply Likes
Hikari -- FWIW, overall, the new cloud update of Feedly seems to meet my needs well enough, although it seems to have a lag in updating articles. (I think I will drop following this discussion now -- too much other work to attend to.)
Photo of alxndr

alxndr

  • 59 Posts
  • 7 Reply Likes
Totally agree -- you've laid out the two big use cases for RSS readers. Personally I only go for the latter, so I don't want anything to disappear ever...
Photo of Samuel Clay

Samuel Clay, Official Rep

  • 6514 Posts
  • 1474 Reply Likes
So by saying it's the training controls is a bit of a misnomer. I can't get rid of the training without completely re-architecting NewsBlur, which won't happen. The real problem is that everybody is going to have different unread counts and they are not only relatively expensive to calculate, but by doubling the number of unreads they will take 2x as long.

The worse problem, and the real reason I haven't launched this yet, is that I store story ids and read stories in redis. It's extremely memory efficient, yet it still takes 18GB of RAM. This is growing at a rate of 2GB a week (on a 32GB machine), and it's the next big piece to try and scale, although I have no clue how I'm going to do it yet.

As soon as I figure out how to store this data more efficiently or in a scalable way, I can finally bump this up to 30 days for everybody and for every feed. But it's going to take a bit of scaling of that redis database to make this happen.

I'll add that if I still cannot figure out how to scale this out for every feed, in about a month I will implement a per-feed unread range, where you'll be able to specify which feeds you want to keep unreads for 30 days. But that'll be a stopgap and a last resort if I can't scale out redis.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
Also, thanks for keeping us updated and giving us some visibility into what you're up against and how hard you are trying.
Photo of romkyns

romkyns

  • 44 Posts
  • 2 Reply Likes
That's interesting to know. Please keep info like this coming, if you can! I followed the NewsBlur development blog hoping to see stuff like this, but it was rather different.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
I agree with romkyns, this was an especially interesting update by Samuel. I would love to see these kinds of updates going forward also.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
Oh, and as for a solution: SAP's HANA super memory resident database would solve your woes (if money was no constraint). :-)
Photo of SirWired

SirWired

  • 5 Posts
  • 0 Reply Likes
If you are chewing up 2GB a week even at 14 day expiration, it sounds like figuring out a solution to this particular scaling problem should be your #1 priority. When you run out of expandability, all the "prettiness" upgrades are going to be kind of moot, because there won't be a NewsBlur.
Photo of chet farmer

chet farmer

  • 3 Posts
  • 1 Reply Like
Again, thank God I didn't pay. Seems like NB is architected for features I (and many others) don't care about at the expense of something we consider basic, i.e. that what's unread should remain unread.
Photo of Livven

Livven

  • 23 Posts
  • 2 Reply Likes
Check out Feedbin. I'm paying for both and while Feedbin certainly is nowhere near NewsBlur in terms of features it tracks unread counts forever, as far as I can tell. I haven't really been using it (heh) but just checked and everything is still unread, including items from February.
Photo of Juan Pablo Ortiz

Juan Pablo Ortiz

  • 9 Posts
  • 7 Reply Likes
Thanks for the tip!
Photo of Henry

Henry

  • 2 Posts
  • 1 Reply Like
As soon as I figure out how to store this data more efficiently or in a scalable way, I can finally bump this up to 30 days for everybody and for every feed.


To be clear, then -- at best, unread items will expire (and become 'read') after thirty days? Not indefinitely?
Photo of John Turnley

John Turnley

  • 1 Post
  • 1 Reply Like
I am a new subscriber. The 14-day (or even a one month) limit changes the way I like to use an RSS reader. It might be something I get used to, or I might find myself using some other service and letting my Newsblur account lapse next year.

I understand the scaling problems and I'm not asking for a magical solution or a refund. I'm just expressing an opinion.
Photo of headsean

headsean

  • 9 Posts
  • 7 Reply Likes
How many unread items would you let stack up in that time? Because there also is a max number of unread stories regardless of the date.
Photo of j.ossevoort

j.ossevoort

  • 7 Posts
  • 0 Reply Likes
Couldn't something like a skip-list be used to handle unreads (or at least unreads longer than 14 days ago)? This is a trick that's is used in one form or the other in NNTP clients and imap servers.

Their base assumption is that most read/unread messages are clustered together (most being read and most unread being groups of consecutive messages). But for them to work you'd need something like a stable incremental ordering of messages (no idea if this fits the newsblur datamodel), like a sequential id or nested intervals for hierarchy storage.
It could always be fixed with another layer of indirection but the costs would possibly be too prohibitive for the newsblur case.
Photo of Daniel Lo Nigro

Daniel Lo Nigro

  • 24 Posts
  • 3 Reply Likes
NewsBlur is open-source so you could take a look at its code and see what it's doing, if you wanted to :)
Photo of yardena

yardena

  • 35 Posts
  • 1 Reply Like
This is an interesting point - read/unread data probably is very compressible and has a lot less entropy than it seems at first glance. But as Samuel said, the issue isn't storage but speed. Still possible with an architecture change, but not trivial, and his bottleneck resource is clearly dev time.
Photo of Christopher Loessl

Christopher Loessl

  • 3 Posts
  • 3 Reply Likes
Why is it a day period a count period anyway? Wouldn't it make much more sense to limit the unread count to a post limit instead? Let's say last 50 posts. So I will never miss a post on low traffic feeds which I usually care more about than the high traffic feeds.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
Good point! Though I would want that to still equate to at least a month on all feeds for most users... Or at least be able to do the equivalent for fairly busy sites like Lifehacker.com and Android Authority.
Photo of Jim Eubanks

Jim Eubanks

  • 7 Posts
  • 6 Reply Likes
This is just my 2-cents. None of the Google Reader replacements IS Google Reader. None of them have had the time or resources Google put into developing and perfecting Google Reader. Time was the essential element that Google didn't allow by giving a relatively short window between announcing and enacting the shutdown. Software takes both time and money to develop. Hardware takes both time and money to configure and scale. NewsBlur, Feedly and many others have done a remarkable job so far in stepping in and successfully tackling a monumental programming task. Expecting perfection at this point from any of these "replacements" is a bit extreme. I have small issues myself, but am overjoyed and thankful, despite these issues, in having such a quality reader available. As, each developer focuses on different strengths and enhancements for their respective readers, different end-users will, of course, find a better fit with one reader versus the next. I would therefore suggest that each person try several replacement readers, as I did, until they find the reader that suits them best. Then, practice the virtue of patience while the hard-working developer works on their behalf to deliver the best reading experience for their user-base.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
Interesting, well-balanced post, Jim.
Photo of Alexey Balmashnov

Alexey Balmashnov

  • 37 Posts
  • 1 Reply Like
Interesting thoughts.

Although talking about issues on how it is difficult to be in time on such a short notice in the topic that is a year old... well it looks to me a bit odd.

Apparently, NewsBlur did a good job in getting more users attracted and converted from Google Reader (would be interesting to see the graph amount of new free/payed accounts in the last days of June).

The big issue here: some of those "converted" have to figure out the limits (or may be better say the differences between the old and the new service) that the service imposes on their habits (and those were formed by a previous service, for sure) the hard way. And it is not difficult to do a bit of expectation management here. Right?
Photo of Nick Kiest

Nick Kiest

  • 15 Posts
  • 2 Reply Likes
I really, really, wish this had been posted somewhere. I likely lost stuff, and did not know it. Rather frustrated.
Photo of Samuel Clay

Samuel Clay, Official Rep

  • 6514 Posts
  • 1474 Reply Likes
You didn't lose anything. All that happens is that unread stories are read. The stories are still there, you just didn't read them for two weeks. That's the way it's always been and will until I finish up the migration that finally gets 30 days working. It's going to take a while until it ships, but I'm working on it.
Photo of Nick Kiest

Nick Kiest

  • 15 Posts
  • 2 Reply Likes
Being marked as read = lost, as far as I am concerned. To me, the primary sync goal is read state. Otherwise I could have 4 independent copies of Net News Wire running, without syncing. To silently change my sync data, after a hidden threshold, is destruction of data.
Glad to hear you are working on it, and I will hang in there, but I am more annoyed about the lack of disclosure, than the technical limitation, at this point.
Photo of Juan Pablo Ortiz

Juan Pablo Ortiz

  • 9 Posts
  • 7 Reply Likes
A ton of people has repeated this, of course is not that he "lost" them as in they "disappeared", but "lost" them in the sense that he didn't even knew they existed and will have to go feed by feed searching for news he thought he read but didn't.
Photo of Benjamin Mako Hill

Benjamin Mako Hill

  • 7 Posts
  • 20 Reply Likes
Samuel, I've seen a half dozen requests to have this documented somewhere like the FAQ since I've been watching this and still no response to that simple and relatively easy to fulfill request. This was yet another request for documenting this feature but you skipped the substance in order to argue with Nick's choice of the word "lost."

75 people have participated in this discussion and almost all of them have been surprised and frustrated. The reason so many people are upset is that there's a gap between their expectations of how NB words and how it actually does. I felt like I had lost material too. You can address this either by changing NB or you can just document it to try to adjust users' (wrong) expectations. What's the objection to an entry in the FAQ?
Photo of PeterB

PeterB

  • 1 Post
  • 0 Reply Likes
To be fair, I didn't realise Google sunset them after 30 days either, and I wouldn't have been happy about that. My most valuable feeds are those with a couple of updates a month, and I got into the habit of logging in every 6-8 weeks and blitzing through all unread items. So I missed items at Google Reader too.

I'd be happy to pay an extra $2-3/month to have my unread figures kept forever.
Photo of Alan D

Alan D

  • 25 Posts
  • 5 Reply Likes
Me too!
Photo of PeterB

PeterB

  • 1 Post
  • 0 Reply Likes
Samuel, I'm getting really intrigued by how to efficiently store unread notifications on such a scale. Googling hasn't turned up much, I clearly haven't found the right terms to search for.

Is this part of the code open source, or private to the hosted-Newsblur only? I plan to have a play and run some simulations on various ideas I have, but I want to see how you're doing it (and the size of the values stored) so I can have an accurate comparison.

You also commented:

The real problem is that everybody is going to have different unread counts and they are not only relatively expensive to calculate, but by doubling the number of unreads they will take 2x as long.

Would accepting an error in the count be acceptable to most people (so you say "20+ unread" etc), and use an algorithm like http://highscalability.com/blog/2012/... ? I know these approaches are more memory efficient, but I don't (yet) know how the computational expense changes.
Photo of happy.cerberus

happy.cerberus

  • 15 Posts
  • 4 Reply Likes
Search for "petascale". It's way beyond what is needed here, but the concept are mostly the same.
Photo of PeterB

PeterB

  • 1 Post
  • 0 Reply Likes
@happy.cerberus Thanks, I tried but it's not the 'scale' term that's the problem. It's what to actually search for. "petascale store unread count" is no more successful than "store unread count" for returning relevant results. Have you any search terms that work?