June 2011 - web-devel

Re: [web-devel] [Haskell-cafe] [ANN] mysql-simple - your go-to package for talking to MySQL
by Johan Tibell 21 Jun '11

21 Jun '11

On Tue, Jun 21, 2011 at 4:47 PM, David Virebayre <dav.vire+haskell(a)gmail.com> wrote: > The problem isn't with the stored procedure, it works if I call it > from the mysql client. Does mysql-simple support stored procedures? Johan

1 0

Yesod IRC meetup
by Greg Weber 21 Jun '11

21 Jun '11

Sunday at 5 UTC. Yesod IRC meetups are focused on working together to make the framework better. We do a great job with our asynchronous communication, but sometimes some quick back-and-forth is a lot more productive. A framework doesn't define a boundary- it is just a collection of libraries to hack on. Is there something you would like to see improved in Yesod? Stop by and get some help hacking on it. See you there! Greg Weber

5 4

Re: [web-devel] [Haskell-cafe] [ANN] mysql-simple - your go-to package for talking to MySQL
by Christopher Done 21 Jun '11

21 Jun '11

On 21 June 2011 16:47, David Virebayre <dav.vire+haskell(a)gmail.com> wrote: > For example, how to write a function that returns the columns of a > table using show columns ? > > type Champ = (String,String,String,String,String,String) > getColumns :: Connection -> String -> IO [Champ] > getColumns conn table = do > query_ conn (fromString $ "show columns from " ++ table) > > if you try query conn "show columns from " ( Only table), the query built is > show columns from 'xxxxx' > which fails. For this you could have newtype Entity = Entity String and implement the Param class for this type which will output the string in double quotes, and then you can write: query conn "SHOW COLUMNS FROM ?" (Only "mytable") (With OverloadedStrings enabled.)

1 1

Re: [web-devel] [Haskell-cafe] [ANN] mysql-simple - your go-to package for talking to MySQL
by Johan Tibell 21 Jun '11

21 Jun '11

On Tue, Jun 21, 2011 at 2:34 PM, cheater cheater <cheater00(a)gmail.com> wrote: > Hi, > does the package adhere to some form of standard API that works the > same way across other similar packages (different mysql drivers, > postgres, mongo, couch, etc)? > > Is there such a standard for haskell? Not at the moment. I believe Bryan has at least talked with one other author (of a PostreSQL binding) about perhaps sharing an API in the future. My opinions is that we should wait to consolidate APIs/standardize interfaces until we actually have an idea what a good Haskell API for databases looks like. To know that we need to seem some different ones. Cheers, Johan

2 1

Re: [web-devel] [Haskell-cafe] [ANN] mysql-simple - your go-to package for talking to MySQL
by Christopher Done 21 Jun '11

21 Jun '11

On 21 June 2011 13:45, David Virebayre <dav.vire+haskell(a)gmail.com> wrote: > The very first example didn't work for me : > ------------------------ > {-# LANGUAGE OverloadedStrings #-} > > import Database.MySQL.Simple > > hello = do > conn <- connect defaultConnectInfo > query conn "select 2 + 2" Yeah, the first example is wrong. Indeed. Should be fixed. I forgot to tell Bryan about that. I noticed it was wrong when I first looked at it but it's still tripping people up. You can access the docs on a slightly earlier version: http://hackage.haskell.org/package/mysql-simple-0.2.2.0 > Next I modified the simple example to call a stored procedure, it > returns a resultset of 12 columns. > Unfortunately, I realised that QueryResults instances are defined up > to 10 elements only. > However, the documentation shows how to define a QueryResults > instance, so I created a datatype and tried to define the instance, > and got stuck with an error: > > Couldn't match expected type `PlateauSel' > against inferred type `Int -> a' > In the expression: convertError fs vs > In the definition of `convertResults': > convertResults fs vs = convertError fs vs > In the instance declaration for `QueryResults PlateauSel' > > Indeed, the documentation shows that convertError takes 3 parameters, > and I gave, as per the example, only 2. > But I'm not sure what to write for the 3rd parameter, the > documentation doesn't help me here. The doc specifies it here: > convertError :: [Field] -> [Maybe ByteString] -> Int -> a > Throw a ConversionFailed exception, indicating a mismatch between the number of columns in the Field and row, and the number in the collection to be converted to. So if you're making an instance for a type that takes ten items from the collection, then put 10. Could always make this clearer. > To try, I put 0, and the test compiled. However, I had a connection > error number 1312, saying my procedure "can't return a result set in > the given context". (The query I used works from the mysql > command-line interface) Ah, I wouldn't know about that, I haven't used the mysql version.

1 0

simple-sendfile
by Kazu Yamamoto 21 Jun '11

21 Jun '11

Hello, I have uploaded the simple-sendfile package to HackageDB. http://hackage.haskell.org/package/simple-sendfile This is a simple version of the sendfile package which use unnecessary system calls (stat() and lseek()). The simple-sendfile package provides native sendfile support for Linux, BSDs, and MacOS. For other OSes, a fallback function using read()/send () is provided. Since the limitation of sendfile on Linux, stat() is called if EntireFile is used. But we can use PartOfFile instead to send entire files. I modified Warp to make use of the simple-sendfile package. https://github.com/kazu-yamamoto/warp/commit/fd73ffb098475255756f7706f83285… And I confirmed that performance becomes much better. I'm very satisfied with the current performance. Since I don't know how T.Handle should be used, the commit above just ignores T.Handle at this moment. If necessary, I would love to modify the API of the simple-sendfile package. --Kazu

2 2

ANN: url-generic-0.1: Parse/format generic key/value URLs from record data types.
by Christopher Done 21 Jun '11

21 Jun '11

Hello, chaps. I thought that it would be an interesting idea to generate URLs from plain old Haskell record data types using Data.Data/Typeable. Repo: https://github.com/chrisdone/url-generic Docs: http://hackage.haskell.org/packages/archive/url-generic/0.1/doc/html/Web-UR… Seems like a success. Whether it's a good idea I'm still considering. Good points: * Works with standard data types. * Derives automatically and therefore predictably. * Disallows conflicts between URL representations (e.g. can't accidentally make up /1234 for "event 1234" and /1234 for "blog post 1234") * The data type, being simple, can therefore be nicely pattern-matched upon, something I use extensively is record wildcarding. * Ordering of parameters doesn't matter. * Fields of type Maybe a can be optionally provided or not. * Enum data types "just work", nice for providing choice in URLs. Bad points: * Sometimes you want custom URLs like "/1234" for convenience. * Not sure how this would interact with web-routes and those type of packages because they use only *one* type for *all* URLs. Let me know your thoughts. Ciao!

3 9

Chris Done on a new pure Haskell pgsql binding and MVC
by Aur Saraf 20 Jun '11

20 Jun '11

None of us should miss this. It especially raised a few questions in my head about the architecture of my site. http://chrisdone.com/posts/2011-06-05-postgres-hpaste-edsls.html

3 5

Yesod IRC meetup summary
by Greg Weber 19 Jun '11

19 Jun '11

I apologize that this is sent twice, but I think only part of it was sent in the previous message. # Groundhog We discussed Persistent and Groundhog with Boris, the author of Groundhog who was kind enough to show up and meet with us. The differences were summarized: * groundhog uses gadts to greatly clean up PersistEntity, and to avoid the need for a bunch of associated types and extra constructors. downside: gadts + type families is a ghc 7 feature, although it might be possible to use phantom types and support ghc6 * persistent requires all datatypes be defines through its QQ syntax, groundhog allows types to be declared as normal. i think this allows persistent to sidestep a lot of sticky naming issues, but groundhog is more flexible * groundhog supports "and" and "or", persistent only supports "and * groundhog use operators, persistent uses constructors * groundhog supports complex arithmetic expressions * groundhog supports comparing fields to fields >From a Persistent standpoint we want all of these features. There are a few minor issues we didn't like, such as preferring our current naming convention to groundhog's, in which 'Field' is appended to every field instead of pre-pending the table name. The main thing we didn't like about Groundhog is its support for Sum data types, which is a neat feature, but complicates the rest of the persistence library, and makes some other functionality more difficult. There are also a few other implementation differences. The direction we want to take Persistent is to incorporate all the good features. This means breaking the current API to make a much nicer one. We plan on providing an upgrade script, much like the upgrade script from hamlet 6 to 7. After Yesod hits 1.0 we will be much more careful to change APIs, all the more reason to make some more drastic changes now. Boris was even kind enough to offer his help in incorporating these features in into Persistent. But 1) he wants to see how it all shakes out first and 2) we don't want to limit his excellent experimentation, so there will be no official merge now. We will continue to try to standardize the APIs of the 2 packages. # Future Meeting Technologies Text meetings can seem a bit laborious at times, although it makes it easy to write this summary. Particularly if we want to hack on some code, it would be much nicer to have audio. I tried out the EVO conferencing tool that was suggested. From a UI standpoint it was a craptacular Java Servlet. It does seem quite featureful, although audio did not work by default on my Ubuntu computer, and the window was annoyingly shaky, possible it doesn't like XMonad. It also uses quite a bit of CPU. So I don't think this is a good solution. I am open to other suggestions. Here are others that I found, but have not tried yet: * https://www.webhuddle.com/ * http://www.openmeetings.de/ Otherwise, we will just do skype breakout sessions as needed. # Future meeting times I created a doodle that people can mark time on next week: http://doodle.com/av92qzmvwyh48hcs Doodle isn't really setup well for large ranges of time though. See you next week! Greg Weber

1 0

RawSql
by Johannes Hess 19 Jun '11

19 Jun '11

Thanks for the cookbook recipe, i made an example where all the rows are extracted and converted back to the original datatype, maybe this is useful for someone else too. {-# LANGUAGE QuasiQuotes, TypeFamilies, GeneralizedNewtypeDeriving, TemplateHaskell, OverloadedStrings #-} import Database.Persist import Database.Persist.Base import Database.Persist.Sqlite import Database.Persist.GenericSql.Raw import Control.Monad.IO.Class (liftIO) import Control.Monad.Loops (whileJust) import Database.Persist.TH import Data.Either (rights) import Data.Text (Text) share [mkPersist, mkMigrate "migrateAll"][persist| Person name Text |] main :: IO () main = withSqliteConn ":memory:" $ runSqlConn $ do runMigration migrateAll insert $ Person "Michael Snoyman" insert $ Person "Miriam Snoyman" insert $ Person "Eliezer Snoyman" insert $ Person "Gavriella Snoyman" insert $ Person "Greg Weber" insert $ Person "Rick Richardson" lol <- withStmt "SELECT name FROM \"Person\";" ([]::[PersistValue]) $ \pop -> whileJust pop (return.fromPersistValues) liftIO $ print (rights lol::[Person])

2 1