2.5.2.0 forgets busy push sources

[grey-hame]

This is becoming particularly irksome: if a non-push source returns "Busy", 2.5.2.0 will wait a while and then try contacting it again; if a push source does so, 2.5.2.0 incorrectly simply forgets the source as if it returned a 404.

As a result, to get a file from a busy push source one must manually keep re-adding the source until the damn thing downloads; else it will not retry at intervals automatically.

This sort of constant manual intervention should not be necessary. If a file exists on the network and you double-click it, it should eventually download without further user intervention, and in the meantime it should not make "negative progress".

Please fix this in 2.5.3.0.

[grey-hame]

Anyone planning to fix this any time soon? Or, at least, acknowledge it?

[old_death]

Have you tried the advanced setting to not automatically forget sources?

[diztrancer]

So Shareaza now have to send all users email "Don't forget to check advanced settings or Shareaza will not work properly" ???

[grey-hame]

[kboddie]

I have noticed this behavior too for a long time, and I also have the appropriate setting checked referenced above.

[old_death]

[diztrancer]

[old_death]

[grey-hame]

I didn't say it sends a 404. I said Shareaza forgets a push source that sends a busy as if it had sent a 404 instead (in particular, Downloads.NeverDrop does not prevent this undesirable behavior).

[grey-hame]

OK. I'm starting to feel like Sisyphus here, constantly re-adding push sources for many of my downloads because they're busy and Shareaza keeps forgetting them.

We've got multiple users confirming this behavior, and it's clearly undesirable.

How soon can we expect to see a 2.5.3.0 that lacks this problem?

[cyko_01]

[grey-hame]

Now I know as much as I did before I asked.

[cyko_01]

[grey-hame]

[ocexyz]

+1
I agree with grey-hame.

I think this is serious bug in Shareaza and should be considered as "critical". However it does not crash Shareaza but just makes it does not download. And more it does disallow Shareaza to share with others Shareazas addresses of known sources as they have been forgotten.

Perhaps fixing this would cause effect present in G1, but not in Shareaza: when I start download from G1 source, then number of sources is increasing in time till max transfer per file is reached, and/or file is downloaded. In G2 I can have only one source and get only from that one, very seldom appears second one, almost never third. If true is that Shareaza can share with other clients addresses of known sources (I suppose yes, it works) and when I have observed Shareaza could not recognise existing source (which I know that was existing one) then logical (for me hehehe of course) is that Shareaza forgets known sources just to soon.

Proposal for consideration. I suppose Shareaza should:
1) remember it (source address) longer then now. Perhaps should be introduced/fixed "do not forget time" like 14 days from last positive confirmation (like download, positive search hit) of "source having needed file" (what is the time of this remembering in G1?)
2) forget source when a hit will return with negative result (source don't offer needed file) but not first one but only perhaps 5th or 7th try? But state of "non existing = not online = none answer, like off line" and "on-line but don't have needed file" should be identified and treated differently
3) time interval between tries (ask for download) should be increasing (this because of possible off line state of source and other user-related behaviour)
4) Those sources which complies conditions in Security rules and filters could be forgotten at once
5) Banned manually sources are forgotten on banning (banned so most probably spam, spam should be forget) "Ban for current session" prevents from accidental killing when dynamic IP are used.

I was talking about above problem with Brov and finally we agreed that this is possibly a bug or bad setting of parameters. Or both.

@grey-hame: however you are right and I support your opinion (+1), could you notice a difference which make my post a bit more easier to accept by developers? They are almost like computers - if you use wrong (not recognised by system) data format then data will be ignored as unreadable (your post will not make expected action=bug fixing). Also good method to use is to make a ticket in bugtrucker, not faster but finally works. Confirmed by experience, believe me... (especially when I have written "not faster" but I don't complain of course )