Wednesday 4 November 2015
SIP/SLIP meeting, October 2015
Thanks to @hamishdickson for compiling these extensive minutes.
Welcomes and apologies
- Welcome to @sjrd who joins us
- @odersky, @jsuereth, @dickwall and @SethTisue are all there
- @dickwall gloats about his holiday
- @non has the flu
- @heathermiller is “otherwise engaged”
1. New Google Calendar
@dickwall has set up a google calendar for slip:
- So far it only has the monthly meetings on there, but that will grow
- It’s public so people can know when the next slip meeting is
- The firstname.lastname@example.org calendar which you can add in google calendars under the “add a friends calendar” option
2. Expert groups
2.1 Getting updates from expert groups
@dickwall can’t email everyone for expert groups, it’s simply not a goer. One idea is to create an event in the new Google calendar (probably the Friday before the meeting) and ask an expert group representative to come to that and provide an update before the next slip meeting.
2.2 Role of expert groups
@dickwall raises a point that was brought up on the IO overhaul discussion, where people feel they have been assigned to an expert group and given work to do without asking. This is entirely not what anyone intended and @dickwall is going to write a blog post about the aim of expert groups, how they are formed and what we would like them to do.
Ideally, an expert group should be self-formed from people with the desire, knowledge and skills to work on a task, then go and do it.
- @dickwall to create a blog post about this issue, he will send it to the Scala elders first so they can voice any opinions
- someone should check if misconstruisem is a word
3. SLIPs without recent activity
Some SLIPs have had no activity in the last couple of months, it would be good if we can somehow track this. @dickwall suggests are create a github tag to track SLIP issues like this. His initial idea is to use “dormant”, but thinks this is too passive-aggressive. People agree and @odersky suggests “hibernate” (which sticks).
4. October milestone issues
Scala Parser Combinators vs FastParse
Link to issue: https://github.com/scala/slip/issues/24
- Parser Combinators is not in the standard library, but is a standard module and while good, it has some issues.
- @lihaoyi has a lib called FastParse (https://github.com/lihaoyi/fastparse), fairly similar in design to Parser Combinators but faster and currently better maintained.
- @odersky has invited @lihaoyi to submit it as a candidate for becoming a standard module
- Currently waiting for @lihaoyi to decide if he wants this to happen and if so to submit a SLIP for it
- Anton Gullett (I can’t find him on github) the maintainer of the Parser Combinators module, should be consulted to see if FastParse meets all the needs met by Parser Combinators.
- @odersky points out that Parser Combinators isn’t a central module, meaning we could actually have both modules for a while
- Note Parser Combinators are mainly community maintained, not primarily maintained by Typesafe or EPFL (obv retain an interest)
- Talk to @gourlaysama about FastParse doing everything required by Parser Combinators
- Invite @lihaoyi to take action if he wants to
Link to issue: https://github.com/scala/slip/issues/23
- @jsuereth spent weekend on @odersky’s strawman proposal (as did Daniel)
- there are a few comments to take care of
- wants to get a fuller proposal out, comments from @DarkDimius (Dmitry)
- wants to flesh out what he has, but it’s on a good path
- collections - @jsuereth wants to do something similar to Java’s collector which is a parallel reduction abstraction as the basis, on top of that have the iterator sequential base collections. @jsuereth needs to spend more time on it, @odersky is keen to see it
- re Views, @odersky thinks we need more data on if it should be iterator based or transducer based - partly why he tried an iterator base
- iterator seems to be more general
- @jsuereth figured out how to do Zip with transducers - required rhs to be an iterator
- @jsuereth’s proposal works on existing collections is fold based, wants to get that in place so we can deprecate the existing set of views asap and focus on longer term collections with less restrictions for view @odersky says that’s reasonable
- @sjrd points out scala-js optimization issues
- @jsuereth and @sjrd to take scala-js optimization issues offline
- @dickwall to create a new issue to track optimization issue
SLIP 22 - Async
Link to issue: https://github.com/scala/slip/issues/22
- @SethTisue is going to ask what @retronym actually wanted/intended to happen with async with regards to the standard library, i.e. if he wanted to include it.
- @dickwall thinks he heard something about issues on Windows, but no one else knew about this.
- @retronym thinks it’s not quite ready for a SLIP yet. He has a round of work planned over the next few months that will include technical improvements to the lib and improved documentation.
- This is on hold for now, but expected to be in better shape in a few months.
Actions: None discussed
Revisit by-name parameters
Link to issue: https://github.com/scala/slip/issues/21
- @dickwall has had no time to look at it
- @dickwall mentions he is wary of being perceived as the SLIP manager and working on issues. Could be seen as a conflict of interest
Actions: None discussed
Scala IO fix/overhaul
Link to issue: https://github.com/scala/slip/issues/19
- This is a long thread. All agree that’s a good thing
- @lihaoyi came up with a good set of questions - first real step in getting this slip going
- @dickwall will reach out to someone and say “can you actually answer these questions?”, maybe @lihaoyi or someone else in the thread
- @dickwall to put all this in a blog post
SIP 24 and 25 - trait parameters and repeating by-names
Link to issue: https://github.com/scala/slip/issues/12
- nothing changed - on hold
Actions: None discussed
Possible SIP for changes to escapes in string interpolation (double quotes inside of a string interpolation)
Link to issue: https://github.com/scala/slip/issues/10
- There has been little movement on this issue mainly for motivational reasons. It sounds like there are varying opinions about what the ideal outcome of this issue is.
- @SethTisue pointed out that escaping quotes in string interpolation is a common newbie FAQ, so there is probably value in completing this SLIP, but exactly what the right change is (if any) is unknown
- @jsuereth pointed out that perhaps we should aim to solve the new person experience rather than solve the whole world’s problems.
- Note, there two similar issues here, one to do with quotes and another to do with backticks
Binary Literals, Underscores in Literals SIP/SLIP tracking
Link to issue: https://github.com/scala/slip/issues/8
- General aim is to see what we can pull in from the Spire library (https://github.com/non/spire)
- Allocate hibernate tag
- Probably an update from @non would be good
Implicits in for comprehensions/pattern matches SIP tracking
Link to issue: https://github.com/scala/slip/issues/6
- @haroldl is still keen on this
- nothing specific heard about progress, give the hibernate tag until more is known
Gather, list, contact interested parties for an Either/Or/Xor/Validation etc SLIP and possible expert group
Link to issue: https://github.com/scala/slip/issues/5
- Either all the way through to validation
- From last meeting, the expert group had an action item of trying out each others libraries. This hasn’t happened due to holidays etc, carry this over
- Expert group to try out each other’s libraries
- collections doesn’t have any tracking - @dickwall to set something up and ping him with any links/info on Gitter and he will collect them and put them into an issue with some links. Expects lots of discussion
- @jsuereth theory: things will die down or start an expert group and actively working on it
- will be fun to analyse trade offs and performance
- issue will have experimentation label
- @jsuereth - call for help, for some sort of source compatibility checker for these libs where you can shove in these new libs and see if things still compile to give us a notion of what we’re breaking
- @dickwall - can we take existing unit tests for collections, clone them off and use them for new collection lib
- @Ichoran (Rex Kerr) has a collection test suite for Typesafe so he’s on that - should be on expert group