August 2011 - web-devel

"Wiki" section of the Yesod doc site
by Michael Snoyman 13 May '12

13 May '12

Hi everyone, I've added a new section to the Yesod doc site: the wiki[1]. This section is intended to be user-writable, so if you have any content you'd like to submit, please go ahead. In particular, this section contains the "Powered by Yesod" page: please add your sites. Unfortunately, the list from the old wiki was lost when my previous VPS died. (All my databases were backed up, I just forgot to backup the Wiki content as well...) Michael [1] http://www.yesodweb.com/page/wiki

5 7

Issues with aeson encoding performance
by Chris Smith 30 Aug '11

30 Aug '11

I know a lot of you use aeson for JSON encoding, and I get the feeling I'm doing something wrong here, so if you can give me some kind of a hint, I'd appreciate it. The aeson package description talks about getting up to around 40K messages per second from the encoding, but I'm pegging the CPU at 100% just trying to send 10 messages per second. Profiling (which is a REAL pain to reproduce, since the GHC API is involved... but I locally butchered a copy of the code enough to route around that) shows JSON serialization taking 98% of the CPU and doing 99.8% of allocations (on the order of 5 GB of total allocation if I leave it running for 15 seconds). Heap profiling also shows the vast majority of the heap occupied by aeson's types (Value, Number, etc), and with huge spikes. The streaming code -- very straightforward; it just repeatedly does a toJSON (getValue (f t)) at most once every 0.1 seconds -- is at https://github.com/cdsmith/gloss-web/blob/master/src/Main.hs#L188 and the serialization (of Gloss's Picture data type) is at https://github.com/cdsmith/gloss-web/blob/master/src/Instances.hs#L106 The code to build the Picture data type is the "animation" function in the second example at: http://cdsmith.wordpress.com/2011/08/20/animations-added-to-web-based-haske… Any ideas? I'm about to try rewriting the JSON serialization to avoid aeson, but before I go there, I thought I'd ask if anyone sees something obviously wrong here. -- Chris Smith

3 4

Handling disconnection in wai or warp
by Hiromi ISHII 29 Aug '11

29 Aug '11

Hi there, I am writing a simple server which returns infinite stream of data to client like Twitter's streaming API. I have to manage the number of clients and the connection times of each one. So I have to handle disconnections during server sending a response to the client. I'm currently considering using wai (+warp) or snap framework. My code fragments for snap and wai are below: ==== snap ver. application ==== stream :: Application () stream = do req <- withRequest $ return . (rqRemoteAddr &&& rqRemotePort) writeLBS $ LBS.unlines $ map (LBS.pack . show) [1..] -- I have to handle disconnection here ==== end ==== ==== wai ver. ==== streamHandler :: Application streamHandler req = do return $ responseLBS statusOK [ ("Content-Type", "application/json") , ("Transfer-Encoding", "chunked") ] (LBS.unlines $ map (LBS.pack . show) [1..]) -- I have to handle disconnection here ==== end ==== Thanks, -- Hiromi ISHII

3 2

Yesod: Dynamic subdomains
by "Timo B. Hübel" 27 Aug '11

27 Aug '11

Hi, I am building a Yesod web app which should support dynamic subdomains. I.e. a user registers as "foobar" on http://www.example.com and can afterwards access her content via http://foobar.example.com Is this possible at all? I am wondering, because the server name is somehow hard coded in Settings.hs as approot. If it is possible: How do I access the domain name in a handler function? Thanks! Timo

3 5

WAI or Snap and the EventSource API
by Mathias Biilmann Christensen 27 Aug '11

27 Aug '11

Hi there, I've been experimenting a bit with the new EventSource API ( http://hacks.mozilla.org/2011/06/a-wall-powered-by-eventsource-and-server-s… ). EventSource looks like a much nicer alternative to WebSockets in all the cases where the connection doesn't really need to be bi-directional (ajax request to push to the server, Event Source to push to the client). It doesn't require a new protocol and should be trivial to pass through HTTP aware load balancers, cache proxies and so on. Have been trying to get either Snap or Warp to serve up an EventSource endpoint, but so far I'm stuck at not being able to get any of them to actually write to the response body before the request is all over. Here's some example code: Snap based: https://gist.github.com/1174102 WAI/Warp based: https://gist.github.com/1174632 The code is mostly the same apart from the glue to the web server. In both cases I'm trying a silly enumerator that gets messages from the keyboard and pass them to the iterator. It all compiles and kinda works, but nothing gets written to the response before I end the enumeration with "q". Any hints on what I'm doing wrong? Or any example code for controlling writes to the response body during a long request? Cheers, Mathias

2 1

wai-logger v0.0.1
by Kazu Yamamoto 25 Aug '11

25 Aug '11

Hello, I have released wai-logger v0.0.1. - This version removes the date thread. Instead new cache mechanism suggested on this ML was implemented. - This version still uses Data.Time. This is because C's strftime did not contribute to improve performance according to my benchmark. - Experimental APIs for Builder are provided: apacheFormatBuilder and hPutBuilder. These APIs are slower but will be revisited when new ByteString library will become available. --Kazu

3 4

[Persistent] PersistBackend and the ErrorT monad
by Max Cantor 24 Aug '11

24 Aug '11

Hi, I am trying to use an ErrorT monad inside some Persist code. I wrote the following instance: newtype OcrError = OE String ... instance PersistBackend m => PersistBackend (ErrorT OcrError m) where replace k v = lift $ replace k v update k l = lift $ update k l updateWhere f u = lift $ updateWhere f u delete = lift . delete insert = lift . insert deleteBy = lift . deleteBy deleteWhere = lift . deleteWhere get = lift . get getBy = lift . getBy count = lift . count selectKeys f = selectKeys f selectEnum f o lim off = selectEnum f o lim off It seems like there is a problem with selectEnum though. In GHCI, this: l :: [PersistKV OcrRefMap] <- rs $ selectEnum [] [] 0 0 works fine, while this: l :: Either OcrError [PersistKV OcrRefMap] <- rs $ runErrorT $ selectEnum [] [] 0 0 hangs. rs is a small function to run persist calls: rs :: (MonadControlIO m) => SqlPersist m a -> m a rs f = do let conn = DT.pack "user=hachicode password=password host=localhost port=5432 dbname=hachicode" withPostgresqlConn conn $ runSqlConn f and you need -XScopedTypeVariables for the ghci commands. Any advice? Thanks, Max

2 2

Re: [web-devel] Caching the System time
by Aristid Breitkreuz 22 Aug '11

22 Aug '11

Did you measure how long it takes to generate that datetime string? Shouldn't that be possible in a hundred nanoseconds or so? Am 08.08.2011 15:01 schrieb "Gregory Collins" <greg(a)gregorycollins.net>:

8 15

wai-logger
by Kazu Yamamoto 22 Aug '11

22 Aug '11

Hello, Sorry for the delay but I have released wai-logger: http://hackage.haskell.org/package/wai-logger I provide low level modules and a high level module. Users should use the low level modules. The high level module is designed for Mighttpd which uses the prefork technique. A sample code is found in its module documentation. Please use this package and give me feedback. And let's consider how to make this as a WAI middle ware. --Kazu

4 8

wai/wai-handler-devel
by Kazu Yamamoto 17 Aug '11

17 Aug '11

Hello, Since I cannot found a way to send a message to yesodweb on GitHub, I'm writhing here. Please advice me if there is another right place. wai/wai-handler-devel has following files: TestApp.hs Testapp.hs testapp.hs I cloned the repository to Mac. Since the default file system of Mac is case-insensitive for file names, the three files are treated as the same and testapp.hs overwrote the others. It would be appreciated if their names will be unique for Mac. --Kazu

2 1