[Haskell-beginners] Options for creating a multiple select form via Yesod
Michael Litchard
michael at schmong.org
Tue Jun 14 19:56:48 CEST 2011
Oh sweet! Thanks!
On Tue, Jun 14, 2011 at 9:43 AM, David McBride <dmcbride at neondsl.com> wrote:
> I think I'll give it a shot. Fun problem and something I'll
> definitely need eventually. I'll try and get you a pull request
> tonight if all goes well.
>
> On Tue, Jun 14, 2011 at 11:40 AM, Michael Snoyman <michael at snoyman.com> wrote:
>> Hi David,
>>
>> Thank you for the analysis, you're absolutely correct that this is a
>> shortcoming in the API of yesod-form as it stands. I had an idea for a
>> possibly simple modification to fix the situation: change fieldParse
>> to
>>
>> [Text] -> Either msg (Maybe a)
>>
>> We don't really want to support returning singles or doubles from the
>> same Field; a multiSelectField will automatically make the "a"
>> variable a list. Said another way, the types should be:
>>
>> selectField :: [(Text, a)] -> Field xml msg a
>> multiSelectField :: [(Text, a)] -> Field xml msg [a]
>>
>> Another related change: we don't really need to separate out
>> fieldRender from fieldView I believe. Instead, we can have fieldView
>> be:
>>
>> fieldView :: Text -- ^ ID
>> -> Text -- ^ name
>> -> a -- currently, this is another Text
>> -> Bool -- ^ required?
>> -> xml
>>
>> I'm willing to test out these changes myself, but wanted to (1) get
>> input from you and (2) see if you or anyone else wanted to take a
>> crack at it.
>>
>> Michael
>>
>> On Tue, Jun 14, 2011 at 4:25 PM, David McBride <dmcbride at neondsl.com> wrote:
>>> I gave a shot at this last night,and didn't quite pull it off. But I
>>> got pretty far and I'd rather you have my work than try it from
>>> scratch.
>>>
>>> The first method won't work because the Field type is used in
>>> validating the get parameters in the mhelper function in
>>> yesod.form.functions. That means that you have to have a unified type
>>> for field that can do everything. Maybe there is a more succinct way
>>> with classes, but I just couldn't think of an elegant way to do it
>>> that way.
>>>
>>> So what I tried to do was make the fieldParser callback into its own type:
>>>
>>> newtype FieldParser msg a = FieldParser (Either
>>> (Maybe Text -> Either msg (Maybe a))
>>> (Maybe [Text] -> Either msg (Maybe [a])))
>>>
>>> That means a field parser can either take one text and return one
>>> item, or it takes a list and returns a list of items.
>>>
>>> Then I went through Yesod.Form.Fields and changed about 12 or 15 references of
>>> { fieldParse = blank $ \s ->
>>> to
>>> { fieldParse = FieldParser . Left $ blank $ \s ->
>>>
>>> Cool, now when you write your multipleSelectField, you'll set the fieldParser to
>>> FieldParser . Right $ etc...
>>>
>>> The very last thing you have to do to fix this is the mhelper
>>> function, which is where I lost steam. Right now it looks up the name
>>> of the field in the get/post params that were passed in, and then
>>> hands the value to fieldParse. What it needs to do is check to see
>>> whether Field Parser is left or right, and then pass in the params
>>> slightly differently depending on which it is. I don't know how the
>>> parameters will end up getting passed into askParams though. Right
>>> now askParams returns a list of names to value pairs, so hopefully you
>>> will end up with a list of multiple entries for the name of your mutli
>>> select and a different value for each entry, which you need to filter
>>> out and collect into a single list and then run the fieldParser on it.
>>>
>>> Hopefully that is not too bad.
>>>
>>>
>>> On Tue, Jun 14, 2011 at 4:30 AM, Michael Litchard <michael at schmong.org> wrote:
>>>> Well I will try the easier way first, and having accomplished that I
>>>> will look into doing it the better way. If people can call dibs, I'd
>>>> like to.
>>>>
>>>> On Mon, Jun 13, 2011 at 9:47 PM, David McBride <dmcbride at neondsl.com> wrote:
>>>>> After looking at the source, you should be aware that
>>>>>
>>>>> 1) yesod-form has been updated to 2.0,
>>>>> 2) it is a lot easier to understand than 1.x was.
>>>>>
>>>>> The main obstacle I see is that the library uses the Field datatype,
>>>>> that has a fieldParse method of Maybe Text -> Either msg (Maybe a).
>>>>> The problem with that is that a multiple select box should require
>>>>> [Text] or perhaps Maybe [Text] rather than Maybe Text. It is making
>>>>> the assumption that there can only be one piece of data per field,
>>>>> which holds for everything except multiple selects and multiple radio
>>>>> buttons.
>>>>>
>>>>> So looking at this, it looks like you'd have to add another field type
>>>>> "FieldMulti" to Yesod/Form/Types.hs, which allows for multiple values.
>>>>> Then add a new version of selectFieldHelper that accepts fieldMultis
>>>>> instead of fields, and then it is trivial to change selectField to be
>>>>> a multi field.
>>>>>
>>>>> Alternatively you could change Field to accept either single or
>>>>> multiple values and change its use everywhere else, which is probably
>>>>> the better answer, but more involved.
>>>>>
>>>>> I don't know if this is the best way to go about it, but it seems like
>>>>> it should work.
>>>>>
>>>>> On Mon, Jun 13, 2011 at 8:11 PM, Michael Litchard <michael at schmong.org> wrote:
>>>>>> Thank you David. I'm trying to figure out step-by-step, exactly how
>>>>>> selectFields binds field values. One thing I'm having trouble with is
>>>>>> visualizing return values.
>>>>>> Beginning with askParams.
>>>>>>
>>>>>> askParams :: Monad m => StateT Ints (ReaderT Env m) Env
>>>>>> askParams = lift askenv <- askParams
>>>>>>
>>>>>>
>>>>>> Here's the example from selectFields
>>>>>> env <- askParams
>>>>>> later on env is used in with the lookup function
>>>>>>
>>>>>> let res = case lookup name env of
>>>>>> seeing as lookup is checking for value of type a in a [(a,b)]
>>>>>> and given the type of askParams
>>>>>> I have no idea what is going on here. I don't see a [(a,b)] in
>>>>>> askParams :: Monad m => StateT Ints (ReaderT Env m) Env.
>>>>>>
>>>>>> So if someone could answer how env <- askParams yields a [(a,b)] for
>>>>>> lookup to use as input, I would appreciate it.
>>>>>>
>>>>>>
>>>>>> On Mon, Jun 13, 2011 at 2:54 PM, David McBride <dmcbride at neondsl.com> wrote:
>>>>>>> The read function is sort of the opposite of the show function. Take
>>>>>>> a string, give me a value. reads is like read, however it has some
>>>>>>> traits that read doesn't have.
>>>>>>>
>>>>>>> The problem with read is that if you go: read "asdf" :: Int, it will
>>>>>>> die with an exception, and that is something you don't want in a web
>>>>>>> app. Also it doesn't tell you what the rest of the string is, so you
>>>>>>> have no real way of finding out what was left of the string after the
>>>>>>> part you wanted to parse.
>>>>>>>
>>>>>>> So there is the reads function that returns [(a,String)] which is a
>>>>>>> list of pairs of the answer a, and the rest of the string String. As
>>>>>>> a bonus, it returns a list so if it can't parse the string you pass
>>>>>>> it, then it just returns an empty list. Why didn't it use Maybe you
>>>>>>> ask? I bet it probably has to do with the function being one of the
>>>>>>> first functions ever written for haskell, long before Maybe existed.
>>>>>>>
>>>>>>> So all it is there is unpack this bytestring into a string, then parse
>>>>>>> it into a value, and please don't blow up if the input is invalid.
>>>>>>>
>>>>>>> On Mon, Jun 13, 2011 at 5:28 PM, Michael Litchard <michael at schmong.org> wrote:
>>>>>>>> I was a bit hasty. I can render a multi-select field easily enough.
>>>>>>>> However, I'm having difficulty following how selectField makes a value
>>>>>>>> from the select field accessible from the handler code calling
>>>>>>>> selectField. Once I figure that out, I can modify multiSelectField
>>>>>>>> accordingly.
>>>>>>>>
>>>>>>>> The goal here being to modify selectField so that a list of field
>>>>>>>> values can be bound .
>>>>>>>>
>>>>>>>> Here's what I have so far:
>>>>>>>> multiSelectField is thus far identical in every way to selectField
>>>>>>>> save for the following change in the Hamlet part.
>>>>>>>>
>>>>>>>> <select multiple="#{theId}" id="#{theId}" name="#{name}">
>>>>>>>>
>>>>>>>> My thinking was that the value bound to multiple was arbitary, and I'd
>>>>>>>> use theId until I figured out something that made more sense.
>>>>>>>>
>>>>>>>> Here's where I am focusing my efforts next
>>>>>>>>
>>>>>>>> http://hpaste.org/47774
>>>>>>>>
>>>>>>>> Specifically
>>>>>>>> (x', _):_ ->
>>>>>>>> case lookup x' pairs' of
>>>>>>>> Nothing -> FormFailure ["Invalid entry"]
>>>>>>>> Just (y, _) -> FormSuccess y
>>>>>>>> I'm thinking this is where selectField binds a value from the select
>>>>>>>> field form. I'm confused by the (x',_):_. At first I thought it meant
>>>>>>>> that just the first pair in a list of pairs is pattern matched
>>>>>>>> against, and the rest discarded. But then I ask myself where the list
>>>>>>>> is coming from. In a select field there would only be one pair, not a
>>>>>>>> list of them. Here's where I get confused. Because if this is not
>>>>>>>> where the values of the select field get bound, I don't know where
>>>>>>>> it's happening.
>>>>>>>>
>>>>>>>> Is my confusion clear enough such that I could get some clarifying
>>>>>>>> feedback? If not, what is unclear?
>>>>>>>>
>>>>>>>> On Sat, Jun 11, 2011 at 11:03 AM, Michael Snoyman <michael at snoyman.com> wrote:
>>>>>>>>> The best way for code contributions in general is to submit a pull
>>>>>>>>> request on Github. If that's a problem, sending a patch via email
>>>>>>>>> works as well (either directly to me or to web-devel).
>>>>>>>>>
>>>>>>>>> Michael
>>>>>>>>>
>>>>>>>>> On Sat, Jun 11, 2011 at 1:14 AM, Michael Litchard <michael at schmong.org> wrote:
>>>>>>>>>> Hey! I just added multiSelectField to the Forms library. I'm only
>>>>>>>>>> getting the first value selected, but I think that's because of how
>>>>>>>>>> I'm using multiSelecrField. I'm going to try to change the client code
>>>>>>>>>> to fix this. I'll let you know how it goes. when I get a
>>>>>>>>>> maybeMultiSelectField added I'll show you what I have. What would be
>>>>>>>>>> the best way to submit this?
>>>>>>>>>>
>>>>>>>>>> On Thu, Jun 9, 2011 at 10:05 PM, Michael Snoyman <michael at snoyman.com> wrote:
>>>>>>>>>>> Hi Michael,
>>>>>>>>>>>
>>>>>>>>>>> There's nothing jQuery or Javascript specific about a multi-select
>>>>>>>>>>> field: it's just a normal select field with a "multiple" attribute. I
>>>>>>>>>>> would recommend taking the selectField code from yesod-form and
>>>>>>>>>>> modifying it to be multi-select. I'll likely do this myself
>>>>>>>>>>> eventually, but it could be a good learning experience in Yesod (and a
>>>>>>>>>>> great introduction to contributing to the framework if you're so
>>>>>>>>>>> inclined).
>>>>>>>>>>>
>>>>>>>>>>> Michael
>>>>>>>>>>>
>>>>>>>>>>> On Thu, Jun 9, 2011 at 8:29 PM, Michael Litchard <michael at schmong.org> wrote:
>>>>>>>>>>>> I'm trying to create a multiple select form, as illustrated on the following:
>>>>>>>>>>>> http://api.jquery.com/selected-selector/
>>>>>>>>>>>>
>>>>>>>>>>>> Here's the options I see possible:
>>>>>>>>>>>>
>>>>>>>>>>>> (1) Write a jQuery widget.
>>>>>>>>>>>> (2) Use plain javascript via Julius
>>>>>>>>>>>> (3) Use the low-level functions in Yesod.Form to write a widget
>>>>>>>>>>>> (4) Use a pre-existing function that does what I need, but am not
>>>>>>>>>>>> aware of this functionality
>>>>>>>>>>>>
>>>>>>>>>>>> (1) has appeal as it looks like something small I can contribute to
>>>>>>>>>>>> the project. It will take me some extra time to figure out the
>>>>>>>>>>>> details. But, I had a look at the other jQuery widgets and they seem
>>>>>>>>>>>> to provide an approachable model to follow.
>>>>>>>>>>>>
>>>>>>>>>>>> (2) This looks like the most straight-forward approach. I'm just
>>>>>>>>>>>> learning javascript so would have to figure out how to capture values
>>>>>>>>>>>> in Haskell from the form.
>>>>>>>>>>>>
>>>>>>>>>>>> (3) This looks like the most difficult way. I don't think I know
>>>>>>>>>>>> enough about the low-level functions in Yesod.Form to be able to
>>>>>>>>>>>> accomplish this in a timely manner.
>>>>>>>>>>>>
>>>>>>>>>>>> (4) This is the best scenario. There's already a way to do this right
>>>>>>>>>>>> now, and I just haven't identified it. If this is the case, I would
>>>>>>>>>>>> appreciate being pointed in the right direction.
>>>>>>>>>>>>
>>>>>>>>>>>> Until informed otherwise, I'm evaluating options 1 and 2. All feedback
>>>>>>>>>>>> welcomed. Thanks to all who made Yesod possible.
>>>>>>>>>>>>
>>>>>>>>>>>> _______________________________________________
>>>>>>>>>>>> Beginners mailing list
>>>>>>>>>>>> Beginners at haskell.org
>>>>>>>>>>>> http://www.haskell.org/mailman/listinfo/beginners
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Beginners mailing list
>>>>>>>> Beginners at haskell.org
>>>>>>>> http://www.haskell.org/mailman/listinfo/beginners
>>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Beginners mailing list
>>>>>>> Beginners at haskell.org
>>>>>>> http://www.haskell.org/mailman/listinfo/beginners
>>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Beginners mailing list
>>>>> Beginners at haskell.org
>>>>> http://www.haskell.org/mailman/listinfo/beginners
>>>>>
>>>>
>>>
>>> _______________________________________________
>>> Beginners mailing list
>>> Beginners at haskell.org
>>> http://www.haskell.org/mailman/listinfo/beginners
>>>
>>
>
> _______________________________________________
> Beginners mailing list
> Beginners at haskell.org
> http://www.haskell.org/mailman/listinfo/beginners
>
More information about the Beginners
mailing list