Posts: 52
Joined: Jun 2009
Reputation:
0
And I am done. The problem isn't in SharpZipLib, it's in ScraperXML.
The HttpRetrieve.GetPage function isn't doing very inconsistent disposing of objects, and it's causing memory to not be reclaimed and is making the web client timeout. Every HttpWebResponse, FileStream, Stream, WebClient, MemoryStream, ZipFile, or StreamReader in that function should be slapped into a using() statement. Do that, and the timeouts go away and you can see the memory usage stabilize (it still oscillates, but it goes down from time to time which it didn't do before).
Is there someplace I can send a patch file?
Posts: 369
Joined: Nov 2006
Reputation:
0
I just realized that if i install the C++ help files in my Visual studio, it's a ton of help to look up all the c++ functions i didn't understand before.
So now there's new update coming soon consisting of almost completely emulated functions of XBMC (encoding is the only thing still yet to be compensated for).
Posts: 369
Joined: Nov 2006
Reputation:
0
Ok http errors hopefully fixed thanks to patch smeehrrr plus i took a few extra steps to ensure memory from Web protocol is fully reclaimed. seems to run alot faster and alot of memory is reclaimed after the function ends.
Will upload later today.
Posts: 52
Joined: Jun 2009
Reputation:
0
Just updated to the latest in SVN and I'm very confused by some of the changes. Can you explain the changes you made to the various MediaTags? It looks like MovieTag no longer has obvious properties like Title and Plot, which has broken my code in many places.
Posts: 52
Joined: Jun 2009
Reputation:
0
More confusion: Why is the Backdrops property on MovieTag a Fanart, but the Backdrops property on TvShowTag is a List<Fanart>?
Posts: 369
Joined: Nov 2006
Reputation:
0
2009-07-24, 23:11
(This post was last modified: 2009-07-24, 23:37 by Nicezia.)
spiff what is the default for input?
it seems a litttle annoying that people keep leaving it off their scrapers as i understood it input could be a string or a buffer reference or an setting replacement.... so if its absolutely nothing? i think if you want input you should have to designate it... but that's just me...
(well actually its only the tvdb that i have this problem with) and of course i can account for it, but its a little annoying because i have to readjust the parser to deal with the scrapers individually (seeing as how the internal flow of XBMC scraper process is still a little confusing to me i have to base my code of what the scrapers do) so now i have like 3 different statements on how to handle the input field, what i'm really wondering is what if someone wants to use blank input (no input value) and buffer $$1 has something in it? does that person have to specify (input="") would that account for a blank input? or would that pull from $$1??
Posts: 369
Joined: Nov 2006
Reputation:
0
going to have to make a major change in the scraper code, so that all scrapers will be managed by scraperxml, the new manner will be that you load a ScraperManager object before calling for a scrape, call this item with the folder that holds scrapers, the path to folder to use for cache. and the Logfile path. This item will hold multiple persistant List<ScraperInfo>(so instead of creating the lists yourself you can just reference them from the ScraperManager) the scrapermanager however doesn't have to be persistant, as when its called it scans the folder for the available scrapers.
it'll be a week or so before this change is made in my code however if the new XBMC code changes before then, i would suggest sticking to the current scrapers that are in the scraperxml svn until those changes are reflected in my code.
Posts: 52
Joined: Jun 2009
Reputation:
0
Can you talk a bit about the justification for this change? What's broken that this will fix?