Thanks everyone. The first suggestion did the trick without needing any other modules.

What instance of Monoid is this?

I dunno, but the context is a gantt chart drawing program: data Task = Task { name :: String, desc :: String, dur :: Days } -- tasks is a list of these and deps is a list of... data Dep = Dep { pre :: String, post :: String } -- referring to Task's name -- draw the squiggles representing sequential dependencies: beziset = foldl atop mempty $ map ((maybe mempty id).bezis) deps bezis (Dep bef aft) = findIndex (\t-> name t == bef) tasks >>= \bti -> findIndex (\t-> name t == aft) tasks >>= \ati -> let t1 = fromIntegral bti in let d1 = finish (tasks !! bti) in --finish is begin + duration let t2 = fromIntegral ati in let d2 = begin (tasks !! ati) in --begin is the latest of the end dates of the directly preceding tasks (or project kickoff) return ( -- the following monstrosity is just about my coordinate system fromSegments [bezier3 (r2 (1.3,0)) (r2 (d2-d1-1,-(t2-t1)*(1+gap))) (r2 (d2-d1,-(t2-t1)*(1+gap))) ] # translate (r2 (descspace+d1,-t1*(1+gap))) ) beziset then gets `atop`ed onto a Diagram. bezis looks far too long as well. Any way to tidy it up? (For some perverse reason I still don't like do notation.) (I know the bezier3 expression is hideous but I can fix that myself.) Adrian. PS: In my previous job I once spent a week evaluating gantt chart drawing softwares. Now I wrote the one I was looking for in half a day. That's Haskell! On 14 May 2013 20:08, Kim-Ee Yeoh wrote:

On Tue, May 14, 2013 at 4:21 PM, Adrian May < adrian.alexander.may@gmail.com> wrote:

...

I have a really annoying scrap of code:

unmaybe Nothing = mempty unmaybe (Just dia) = dia

It happened because I'm using Diagrams but building my diagram requires looking something up in a list using findIndex, which returns Maybe Int.

What instance of Monoid is this? Because Int has both a Sum Int and a Product Int instance so you can't just apply unmaybe to (Just 3 :: Maybe Int).

Defining unmaybe Nothing = 0 prompts the question: how will you distinguish misses versus hits on the head of the list? Presumably you don't want to.

You might be interested in the totalized lookup functions defined in my private toolkit (hayoo returns nothing):

-- tlookup :: (Eq a) => b -> a -> [(a, b)] -> b tlookup b a abs = fromMaybe b $ lookup a abs tlookup0 a abs = tlookup mempty a abs

-- Kim-Ee

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

...

For some perverse reason I still don't like do notation.

You're in fine company. Many of the senior folk shun it altogether, especially in a classroom setting. See [1].

Well that's nice to know. I feel that I can see what's going on better in the non-do notation, but when I first started C I used to prefer *(p+i) to p[i] quoting the same rationale. That seems silly to me now.

...

bezis looks far too long as well. Any way to tidy it up?

The list indices ati and bti are used only once: to retrieve the (first) item matching the predicate. Might Data.List.find be a better fit?

No, I use them via t1 and t2 to figure out which row to put the squiggle on. The attached picture might put it all in perspective.

Replacing findIndex with find should get rid of half of the lets that are just fromIntegral noise.

Yeah but I still need the row number.

This sort of falls under the broader rubric of more whitespace is better than less, all other things equal. So consecutive

...

let ... in let ... in

can be merged into a single multiline let, pace C&P diehards.

True. Thanks, Adrian.

And finally to answer the earlier question: fromSegments returns a PathLike-constrained value. PathLike is subclassed from Monoid.

[1] http://www.haskell.org/haskellwiki/Do_notation_considered_harmful

-- Kim-Ee

On Tue, May 14, 2013 at 9:56 PM, Adrian May < adrian.alexander.may@gmail.com> wrote:

...

Thanks everyone. The first suggestion did the trick without needing any other modules.

...

What instance of Monoid is this?

I dunno, but the context is a gantt chart drawing program:

data Task = Task { name :: String, desc :: String, dur :: Days } -- tasks is a list of these and deps is a list of... data Dep = Dep { pre :: String, post :: String } -- referring to Task's name -- draw the squiggles representing sequential dependencies: beziset = foldl atop mempty $ map ((maybe mempty id).bezis) deps bezis (Dep bef aft) = findIndex (\t-> name t == bef) tasks >>= \bti -> findIndex (\t-> name t == aft) tasks >>= \ati -> let t1 = fromIntegral bti in let d1 = finish (tasks !! bti) in --finish is begin + duration let t2 = fromIntegral ati in let d2 = begin (tasks !! ati) in --begin is the latest of the end dates of the directly preceding tasks (or project kickoff) return ( -- the following monstrosity is just about my coordinate system fromSegments [bezier3 (r2 (1.3,0)) (r2 (d2-d1-1,-(t2-t1)*(1+gap))) (r2 (d2-d1,-(t2-t1)*(1+gap))) ] # translate (r2 (descspace+d1,-t1*(1+gap))) )

beziset then gets `atop`ed onto a Diagram.

bezis looks far too long as well. Any way to tidy it up? (For some perverse reason I still don't like do notation.) (I know the bezier3 expression is hideous but I can fix that myself.)

Adrian.

PS: In my previous job I once spent a week evaluating gantt chart drawing softwares. Now I wrote the one I was looking for in half a day. That's Haskell!

On 14 May 2013 20:08, Kim-Ee Yeoh wrote:

...

On Tue, May 14, 2013 at 4:21 PM, Adrian May < adrian.alexander.may@gmail.com> wrote:

...

I have a really annoying scrap of code:

unmaybe Nothing = mempty unmaybe (Just dia) = dia

It happened because I'm using Diagrams but building my diagram requires looking something up in a list using findIndex, which returns Maybe Int.

What instance of Monoid is this? Because Int has both a Sum Int and a Product Int instance so you can't just apply unmaybe to (Just 3 :: Maybe Int).

Defining unmaybe Nothing = 0 prompts the question: how will you distinguish misses versus hits on the head of the list? Presumably you don't want to.

You might be interested in the totalized lookup functions defined in my private toolkit (hayoo returns nothing):

-- tlookup :: (Eq a) => b -> a -> [(a, b)] -> b tlookup b a abs = fromMaybe b $ lookup a abs tlookup0 a abs = tlookup mempty a abs

-- Kim-Ee

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

On Wed, May 15, 2013 at 9:58 AM, Adrian May wrote:

The list indices ati and bti are used only once: to retrieve the (first)

...

item matching the predicate. Might Data.List.find be a better fit?

No, I use them via t1 and t2 to figure out which row to put the squiggle on. The attached picture might put it all in perspective.

...

D'oh. Didn't see the later references, sorry.

Something to think about is: how would you document "my coordinate system" as you wrote in your comment? Highly visual things like these are challenging to describe even for verbal-thinking types. Probably sticking the picture in png/pdf format and linking to it in the comments is a good a solution as any. -- Kim-Ee

On Wed, May 15, 2013 at 10:37:56PM +0800, Adrian May wrote:

I'd best give names to those forks though otherwise the whole project plan will break whenever I insert a new one. Can I give a parameter to an operator? Like:

task1 `fork "name"` taskb

task1 ( >^> "name" ) taskb

No, unfortunately neither of those is possible. -Brent