Feeds with articles older than FORCE_ARTICLE_PURGE come back in as unread?

[nneul]

It appears that if you have a FORCE_ARTICLE_PURGE setting, and the feeds have articles coming in that have dates older than that setting - they are brough back into the system as unread.

I've not confirmed this at low level, but it seems like if that is set, the updater should probably discard/ignore "new" articles that would otherwise be excluded by that.

If this isn't the case, then I have a weird symptom where certain feeds (ones that have really old articles in the feed along with new ones) are regularly showing up with unread articles that have long since been marked read.

Suggestions?

[fox]

this isn't how this works. purging is done after feed was updated and goes by internal timestamp which is bumped every time an article is encountered in the feed.

if the aforementioned option is set, a fixed interval is used (and purging unread articles is enabled) nothing else changes.

my only suggestion is that you managed to set the interval so low it somehow glitches out in which case i suggest you set it a bit higher.

also, read the rules before posting here any further.

[nneul]

ttrss: 63f0ed3
running on ubuntu 16.04 current, apache, mysql percona 5.6.35, php 7.0.15
No errors in updater output or apache.

Would you consider "30 days" to be a interval that is too low? The only reason I even have this on is I has gotten tired of seeing ancient articles popping back into my unread list in certain feeds and thought it might resolve that.

I'll capture a feed debug the next time I see one of the bogus/excessively old articles. (talking about stuff where articles from last year and prior years coming back in the unread section of the feed.) It's happening on a fairly regular basis on certain feeds. Note that they don't seem to show any problems with the feed when passed through myfeedsucks.

[fox]

> The only reason I even have this on is I has gotten tired of seeing ancient articles popping back into my unread list in certain feeds and thought it might resolve that.

this has nothing whatsoever to do with that and is not going to help at all

the only way for "old" articles to pop up again as unread if feed explicitly changes GUIDs for some reason i.e. the feed is broken, no amount purging settings is going to fix that, everything else should be handled automatically the way i outlined in the above post

[nneul]

So it happened again today. I took a look at the feed debug - and what I've been able to determine from comparing this to previous (yesterday) backups is that the (arbitrarily picked) article was not present in the older database at all. This same feed showed up with articles going back for several years.

The feed itself only has about 50 articles in it, but around 2043 entries in database for this particular feed.

Is there any way to ask the updater to retain a copy for debugging of every feed it retrieves?

From a look at the raw feed though - and the apparent randomness of the dates - I suspect that the upstream feed is just garbage. (Well, we already know that it's drivel, but still entertaining.)


Feed URL: http://feeds.feedburner.com/Failbooking

Code: Select all

[14:44:05/535] article processed
[14:44:05/535] guid 2,http://cheezburger.com/1085957/twitter-onlyacceptabletoday-list-there-are-some-things-that-are-only-okay-to-do-on-halloween / SHA1:000c8d6e7aa3c88c78f3aac2d7416f7709b798e6
[14:44:05/535] orig date: 1477940400
[14:44:05/535] date 1477940400 [2016/10/31 19:00:00]
[14:44:05/535] title Twitter Shared With the Hashtag #OnlyAcceptableToday That There Are Some Things That Are Only Okay to Do on Halloween
[14:44:05/535] link http://cheezburger.com/1085957/twitter-onlyacceptabletoday-list-there-are-some-things-that-are-only-okay-to-do-on-halloween
[14:44:05/535] author
[14:44:05/535] num_comments: 0
[14:44:05/535] looking for tags...
[14:44:05/535] tags found: twitter,list,halloween
[14:44:05/535] done collecting data.
[14:44:05/535] article hash: de0c88298c07a47bb6e5c5e4154146aaf59d7faf [stored=de0c88298c07a47bb6e5c5e4154146aaf59d7faf]
[14:44:05/535] hash differs, applying plugin filters:
[14:44:05/535] ... Af_Comics
[14:44:05/535] === 0.0000 (sec)
[14:44:05/535] ... Af_Unburn
[14:44:05/535] === 0.0000 (sec)
[14:44:05/535] plugin data: af_comics,af_unburn,
[14:44:05/535] matched filter rules:
[14:44:05/535] filter actions:
[14:44:05/535] article labels:
[14:44:05/535] force catchup:
[14:44:05/535] base guid found, checking for user record
[14:44:05/535] initial score: 0 [including plugin modifier: 0]
[14:44:05/535] user record FOUND
[14:44:05/535] RID: 460565, IID: 461616
[14:44:05/535] assigning labels [other]...
[14:44:05/535] assigning labels [filters]...
[14:44:05/535] looking for enclosures...
[14:44:05/535] article enclosures:
Array
(
)
[14:44:05/535] filtered article tags:
Array
(
    [0] => twitter
    [1] => list
    [2] => halloween
)
[14:44:05/535] article processed


[fox]

>Is there any way to ask the updater to retain a copy for debugging of every feed it retrieves?

you can log to file, i think, but i'm not sure how it would work with debugging