@dwmatiz Currently we have indirect control over the cache size: via number of messages to store and number of days to store them. I don't think we need to restrict database size separately, but I agree that we need this additional setting: "Maximum size of cached images, Mb" with a default value of 100 Mb ?! I.e. we may decide to store messages for years, but have a reasonable limit for avatars and attachments.
And here is how that old discussion on #SocialNetwork term looks like now in #AndStatus (I like the dark theme although light theme is a default one now...) @maloki
@KARiley40 Try #AndStatushttps://play.google.com/store/apps/details?id=org.andstatus.app Maybe it's not as pretty looking as others, but it will allow you to try other Social networks also, and it works offline... You can even use it as a tool to explore Social networks, joining Search query results from all networks in a single view....
@perspicacious01 On a positive side - it's (at least mostly) Open Source software, so GNU Social could easily borrow something good from it?! I didn't inspect #Mastodon's server infrastructure, but from what I've heard, distribution the preconfigured Social network instance as a #Docker image could simplify adoption / upgrades of the sever...
@archaeme I feel that "equality of service principle" would actually make #Mastodon much like #Twitter: It's also "decentralized" in some sense, because it consists of a multitude of servers distributed globally. Also, according to some regional laws , personal data of a country citizens should be physically stored on a territory of the country: here we see real data decentralization...
@colegota Please set logging level to VERBOSE. What I cannot understand is: what data is being transferred? I mean if a server is down, it doesn't respond. By default, in #AndStatus failed commands are repeated not more than 10 times, and not less than after several minutes after previous attempt...
@musicman I also noticed that sometimes older messages are being downloaded after newer. Maybe I observe this, because I use several accounts so sometimes I receive the same messages via different channels in different times. As you may have noticed, #AndStatus collapses duplicated posts in one, so sometimes "new" messages are even not shown, they only add "+1" to an existing message's footer...
@hector I cannot reproduce an issue in my tests. Could you point me to exact message or at least to exact user, whose username "breaks AndStatus"?! And what exactly "breaks", please describe.
@punk That GNU Social client's protocol is usually called "Twitter compatible" (old Twitter REST API v.1.0) ?! But I'm pleased that #AndStatus becomes a common noun :-) @paulfree14
@hector Thank you very much for the report! It's really a please to receive reports about real bugs :-) I will look at this in 12 hours, so you have time to find some more bugs to be fixed today :-) Please note that the latest production version of #AndStatus is 33.07.
Yuri Volkov (yvolk@loadaverage.org)'s status on Sunday, 14-May-2017 07:32:14 UTC
Yuri VolkovYesterday I was implementing #Mastodon API in #AndStatus. Comparing AndStatus with #Tusky client at some point I noticed that Tusky shows me a new post, which mentions myself. But I couldn't find the same post in any timeline that AndStatus feeds from the Mastodon server. It appeared that in addition to Twitter/GNU Social-like timelines Mastodon has something conceptually similar to an "Activity stream" that is called "Notifications" here and, as other parts of the API, is "hand made" :-) Actually, this is a good sign, showing that ideas of #ActivityStreams (currently represented by the #ActivityPub standard https://www.w3.org/TR/activitypub/ ) are growing in the heads of the Mastodon developers. I would go one step further and formulate all timelines and actions as activities and activity streams... It's interesting that these ideas are growing in _my_ head also :-) - I decided that internal AndStatus model of notifications will be based on ActivityStreams too. I.e. each individual notification will have an Actor, an Action and an Object/Content e.g. "Peter posted 'I met a friend'". Having notifications in such a structure, we will be able to show them to a User in any needed form: in the Android Notification bar or as a timeline. Currently AndStatus shows mostly "mentions" only in its "Mentions" timeline...
>> Aqeel Zafar: “[...] I rarely have anything interesting to share, so [...]” Oh, but you can contribute a lot just by commenting on posts or sharing them =) Development of the Pump.io core picked up pace over a year ago. Other parts never stopped, actually. We're actually on the way to adopting the new ActivityPub standard, which will be adopted by other federated networks too. That will allow Pump.io to interfederate with those other players in a big network. -- jankusanagi@identi.ca URL: https://identi.ca/jankusanagi/comment/F0MY_vUqTCqaxTFTjpSahQ
@nestort In short: As a developer I don't have any means to check this. #F-Droid has some automated procedure that monitors a source code repository (Github in this case) and figures out, when a new release should be built. Then it automatically builds the application from its source code and makes it available to F-Droid users. This process can fail at any step, and I don't have any "support person" to check and fix this. See our discussion of the similar case of broken process with F-droid team member: https://github.com/andstatus/andstatus/issues/326
@perspicacious01 As I understand, full text search would require much more resources than a simple hashtag indexing. On the other hand, knowing that hashtags are searchable, allows to write messages with predictable #discoverability. I mean that adding a known hashtag to a message will ensure that the message will be present in the "Hashtag timeline". BTW in #AndStatus such "Hashtag timelines" (like other Search timelines) are the "first class timelines". I.e. you can sync and use them exactly like e.g. Home Timeline.