Google Reader Sync

[barijaona]

A new beta is available at github.com (https://github.com/ViennaRSS/vienna-rss/downloads).

Thanks to Andrew Herron (aka TheSpyder) for his contributions !
I am eager to have your feedbacks, especially those from Shawn and Ben.

[David Munch]

How stable is it by now?

[shawn]

Still a non-starter for me, unfortunately.

[benst]

A new beta is available at github.com (https://github.com/ViennaRSS/vienna-rss/downloads).

Thanks to Andrew Herron (aka TheSpyder) for his contributions !
I am eager to have your feedbacks, especially those from Shawn and Ben.

Hi Barijaona,

It worked for a day or two, and I noticed the following:

- I can rearrange the order of the feeds now, great!
- If I have multiple feeds inside a folder, and I click on the folder, I see all messages. Then I hit K to mark all as read. The message list correctly shows them as read. However, when you go to a different folder and back again, the messages are again displayed as unread.
- Nothing changed here, but now suddenly Vienna crashes on startup. Crash log attached.

Kind regards,
Ben

[barijaona]

A new beta (build 2804) is available at Github https://github.com/ViennaRSS/vienna-rss/downloads.

Most of the effort has been on stability, which is particularly tricky when one uses multi-threading...
Your feedbacks are welcome.

[benst]

Unfortunately also crashes on startup. Crash log attached.

Ben

[shawn]

Ditto…

Do you have an idea of where this may stem from? I get the spinning beach ball of doom for a few seconds before the crash actually hits. If you know about where to look, I can build from here and poke around, myself.

[barijaona]

Hello everybody,

I have attached the dSYM herewith :

ViennaBeta.app.dSYM.zip

dSYM file for ViennaBeta build 2804

(2.81 MiB) Downloaded 467 times

Through the atos command, one can see that most crashes involve the following call stack (listed from innermost call to outmost) :

-[Database initArticleArray:] (in ViennaBeta) (Database.m:1972)
-[Database createArticle:article:guidHistory:] (in ViennaBeta) (Database.m:1417)
-[GoogleReader feedRequestDone:] (in ViennaBeta) (GoogleReader.m:259)
-[ASIHTTPRequest reportFinished] (in ViennaBeta) (ASIHTTPRequest.m:2028)

Sometimes, instead of having line 259 of GoogleReader.m involved, one has line 890 of RefreshManager.m, which is rather similar...

This issue clearly involves multithreading. I experienced a much more stable behavior replacing this section at line 715 of RefreshManager.m

Code: Select all

		[self performSelectorInBackground:@selector(finalizeFolderRefresh:) withObject:[NSDictionary dictionaryWithObjectsAndKeys:
																					folder, @"folder", 
																					connectorItem, @"log", 
																					url, @"url",
																					receivedData, @"data",
																					lastModifiedString, @"lastModifiedString",
																					nil]];

with :

Code: Select all

		[self performSelectorOnMainThread:@selector(finalizeFolderRefresh:) withObject:[NSDictionary dictionaryWithObjectsAndKeys:
																					folder, @"folder", 
																					connectorItem, @"log", 
																					url, @"url",
																					receivedData, @"data",
																					lastModifiedString, @"lastModifiedString",
																					nil]
																					waitUntilDone:YES];

I hoped fixing the isRefreshingFolder method in RefreshManager should improve things, but had a limited success so far...

Some problems may also be caused by methods returning a NSArray or a NSMutableArray or a NSDictionary, like guidHistoryForFolderId in Database.m ... The implementation may create memory troubles, as discussed here http://stackoverflow.com/questions/2488 ... n-an-array. But once again, my improvement attempts have been unsuccessful...

Your ideas are welcomed... and if we are unable to understand the multithreading problem, there is still the crude solution above...

[barijaona]

I personally experienced that some of these problems might be related to a database corruption... Unfortunately, it is quite possible : we are still at a beta stage, aren't we ?

@shawn and @benst : would you be kind enough to check if the problems reoccur when restarting from a fresh database ? (put aside the ~/Library/Application Support/Vienna folder) (and export your subscription list first...)

Regards.

[barijaona]

The latest build, that I uploaded today and is numbered 2806, seems more stable to me.

But your mileage may vary...

[shawn]

Weird…I've attached crash info from starting up after removing the entire ~/Library/Application Support/Vienna folder, and everything I found for it in ~/Library/Preferences.

[barijaona]

@shawn, in ~/Library/Preferences/uk.co.opencommunity.vienna2.plist , do you have a FilterMode key ?

What value is associated ?

[phloo]

Just installed ViennaBeta and tried to test it.
Anything I should do before using it? Cause its crashing...

[barijaona]

Thanks for the reports.
Apparently, problems can occur when Vienna is trying to update a view which is currently refreshing.

It should be fixed soon.

[phloo]

It should be fixed soon.

Glad to hear that! Very appreciated..

[barijaona]

The build numbered 2807 should fix the problem reported by phloo and shawn.

[phloo]

it should but it didnt. still crashing.
and the refresh icon never stops turning.

[shawn]

@shawn, in ~/Library/Preferences/uk.co.opencommunity.vienna2.plist , do you have a FilterMode key ?

What value is associated ?

No FilterMode key. 2807 started up without crashing! Poking around to check things out and get it back to how I like it.

[barijaona]

Build 2808 is out at https://github.com/ViennaRSS/vienna-rss/downloads

It brings some bugfixes.

[benst]

2808 still crashed for me on startup, see attached crash report.

After I have removed the ~/Lib.../Vienna directory it starts up ok, so I'll give it a go the next few days and see what happens.

Ben