March 2011 - web-devel

Web framework comparison
by Zachary Kessin 20 Mar '11

20 Mar '11

I created a spread sheet<https://spreadsheets.google.com/ccc?key=0AscGbs8KnSyAdGNwSnp1LWZUTlJyZUxNaF…> to compare the features of the major Haskell web frameworks. Currently It has Hasckstack, Yesod and Snap (but feel free to add others). I listed a bunch of the categories that are important to me, but it is is far from a complete list. So I am sharing this document with the community to let others fill in what they think is important. --Zach

2 1

Interpolating url does not work
by vagif.verdi＠gmail.com 20 Mar '11

20 Mar '11

I have 2 handles defined in a separate module. Both handles are in the same module. When i interpolate one into the template of the other i get an error "Not in scope". But both handles work ok, If i do not interpolate, i can hit manually both of them. Handle denied as getFormR :: Int -> Handler RepHtml I interpolate it like this: <a href=@{FormR 11}> and get error: Not in scope: data constructor `FormR' Most likely i am missing something obvious. But cannot figure it out. Regards, Vagif Verdi

2 2

Need a subsite example/tutorial/documentation
by vagif.verdi＠gmail.com 18 Mar '11

18 Mar '11

I am trying to build an intranet web app that would need access control (different modules are allowed/denied to different groups of users). I am thinking about putting a handle that checks credentials and then lets routing to continue to match the rest of the route. /app/moudle1 <- Here i check the credentials /app/moudle1/path1 <- If credential are OK, then the handler falls in here. /app/moudle1/path2 How do i do that with yesod ? Is it something that needs a subsite ? I would appreciate a small example. It would help me a lot. ----------------- Another question, Is there a way a automatically and transparently for developer encrypt part (or all ) of the url. Like this: www.website.com/app/jshdfbjkshabfjhvkjvjaksvdfjhvasjdhvfkjshadfhjasdhfkjasf… where the encrypted part will be converted to normal route by the rounting handler, and when urls are generated, encrypted back. Of course using a separate key for each user/session. Regards, Vagif Verdi

4 7

Upcoming WAI changes
by Michael Snoyman 16 Mar '11

16 Mar '11

Hi all, I wanted to bring up for discussion some planned changes to the WAI. You can see what I'm proposing on Github[1]. The first thing you might notice is that the Network.Wai module is significantly smaller than it used to be. This is because a number of the constants and data types previously defined in WAI are now provided by the http-types[2] package. Besides the smaller code base, the following changes are also included: * HttpVersion is no longer a ByteString; instead, it is a datatype with two Ints. So HTTP/1.1 becomes HttpVersion 1 1. * pathInfo is renamed to pathInfoRaw, and the new pathInfo is a [Text] value. This represents the *parsed* path info. For example: GET /foo/bar HTTP/1.0 becomes ["foo", "bar"] By including this in the definition of WAI, it's simpler to provide sub-apps. For example: masterApp req = case pathInfo req of "site1":rest -> site1 $ req { pathInfo = req } ... * Similarly, queryString becomes queryStringRaw and queryString is now a [(ByteString, Maybe ByteString)]. This allows the query string to be processed once and the value shared amongst all middlewares and applications. If you are wondering why we use ByteString and not Text, it is because the query string can have any arbitrary encoding. The path info, on the other hand, is (basically) required to be UTF-8. * In place of CIByteString, we now have CIAscii. Besides the slightly nicer name, this is now taking advantage of the newly written ascii[3] and case-insensitive[4] packages. * Header values are given as Ascii values instead of ByteStrings. If anyone has concerns about these changes, please raise them. I'd rather get all the complaints out in the open before releasing a new version. Michael [1] https://github.com/snoyberg/wai/blob/master/Network/Wai.hs [2] http://hackage.haskell.org/package/http-types [3] http://hackage.haskell.org/package/ascii [4] http://hackage.haskell.org/package/case-insensitive

2 3

json to string/bytestring?
by Max Cantor 16 Mar '11

16 Mar '11

Is there a preferred and simple way to get a string out of a json value? Max

2 1

Persistent: IdIn constructor added
by Michael Snoyman 15 Mar '11

15 Mar '11

Hi all, I've added the recently discussed IdIn constructor to the Persistent Github repo. I have a test case which is passing, but I wouldn't mind some independent observations before proceeding with a release. Michael

1 0

Yesod. Hamlet problems
by Anton Cheshkov 15 Mar '11

15 Mar '11

Hello Michael. Hello All. Please help me to explain what is wrong i have a function which worked before i add line *<img src="StaticR p1_png"> * : *toHamlet :: [SearchResult] -> String - > Int -> Html* *toHamlet results queryId n = [$hamlet|* *$forall r <- results* * <div #search-result-block qid="#{queryId}" piece="#{show n}">* * <div .image-block .left>* * <img src="StaticR p1_png">* * <div .info1-block .left>* * <span>#{snd $ snd r}* * <div .info2-block .left>* * <div .clear>* *|]* I have the compile time error: *Couldn't match expected type `hamlet-0.7.1:Text.Hamlet.Quasi.R:HamletUrlHtmlM'* * against inferred type `EstateRoute'* * In the first argument of `\ u[ahZb]* * -> hamlet-0.7.1:Text.Hamlet.Quasi.urlToHamletMonad* * u[ahZb] []', namely* * `StaticR p1_png'* * In a stmt of a 'do' expression:* * \ u[ahZb]* * -> hamlet-0.7.1:Text.Hamlet.Quasi.urlToHamletMonad u[ahZb] []* * (StaticR p1_png)* * In the expression:* * do { (hamlet-0.7.1:Text.Hamlet.Quasi.htmlToHamletMonad* * . preEscapedString)* * "<div id=\"search-result-block\" qid=\"";* * (hamlet-0.7.1:Text.Hamlet.Quasi.htmlToHamletMonad* * . Text.Blaze.toHtml)* * queryId;* * (hamlet-0.7.1:Text.Hamlet.Quasi.htmlToHamletMonad* * . preEscapedString)* * "\" piece=\"";* * (hamlet-0.7.1:Text.Hamlet.Quasi.htmlToHamletMonad* * . Text.Blaze.toHtml)* * (show n);* * .... }* I do not understand why (. Thanks. -- Best regards, Cheshkov Anton Phone: +7 909 005 18 82 Skype: cheshkov_anton

2 1

Python web benchmark help requested
by Michael Snoyman 14 Mar '11

14 Mar '11

Hi all, The Yesod team is in the middle of getting some benchmarks up coming Warp to web servers in other languages. Unfortunately, no one on the team is enough of an expert in Python to come up with a good benchmark. Is anyone out there familiar with the Python web scene, and interested in lending a hand? We're running all the tests on an EC2 instance; I can give you access to set things up however you think is best. Thanks, Michael

3 2

Re: [web-devel] [Yesod][Persistent] FooIdEq,FooIdIn
by Michael Snoyman 14 Mar '11

14 Mar '11

Well, let's discuss the motivations for requiring the annotations at all. I *think* no one but me really knows why I did this, because I have neither documented my intentions nor taken advantage of them code wise. While it's true that it lowers the number of constructors created at compile time, that was *not* my primary concern. The idea was to allow for optimizations: for a SQL backend, the SQL migration code could produce indices automatically. For other backends, it could have an even more dramatic effect on how the data is stored. But I think that argument doesn't really apply to an IdIn constructor. Greg, I like your proposal of including it by default. If there are no objections, let's try to get it in for the next release. One other note about the next Persistent release: we're going to be splitting the package up into a core and a Template Haskell set. Some people are using Persistent in an iPhone app, and the iPhone compiler cannot support TH. Just a little heads-up. Michael On Sat, Mar 12, 2011 at 9:27 PM, Greg Weber <greg(a)gregweber.info> wrote: > It might be better to save the space after the table line just for future > table annotations. > Then to annotate id, it would be added as a field. > This brings up another point though- do we need to add "In" annotation- why > not just enable it? Does it make the compiles take really long, etc? We > should at least document the rational in the persistent section. > On Sat, Mar 12, 2011 at 10:46 AM, Michael Snoyman <michael(a)snoyman.com> > wrote: >> >> On Sat, Mar 12, 2011 at 3:07 AM, Katsutoshi Itoh <cutsea110(a)gmail.com> >> wrote: >> > Hi >> > >> > I had often wanted FooIdIn, FooIdEq on Foo Persistent type. >> > Would you support this by default? >> > >> > from japan. >> >> As Max said, glad to see that you're safe. >> >> For FooIdEq: I don't think there is any need, since it can be >> expressed more succinctly with the "get" function. As for FooIdIn: >> this *was* a recently requested feature. Maybe we can get it in for >> the next release. I think it would make sense to add the In attribute >> to the entity itself, eg: >> >> mkPersist [$persist| >> Person In >> name String >> age Int >> |] >> >> Any objections? >> >> Michael >> >> _______________________________________________ >> web-devel mailing list >> web-devel(a)haskell.org >> http://www.haskell.org/mailman/listinfo/web-devel > >

1 0

Yesod. Persistent. Disjunction
by Anton Cheshkov 13 Mar '11

13 Mar '11

Hi Michael. Please show the way how to perform a query for persistent in disjunction form. For example select * from Table t where t.c1 = 1 or t.с1 = 2 Now, for example, selectList function applies filters in conjunctive. After that it would be great to see how to make query like select * from Table t where t.c1 in [1, 2] Thanks. -- Best regards, Cheshkov Anton Phone: +7 909 005 18 82 Skype: cheshkov_anton

3 2