Stackage: megaparsec 6.0

[snoyberg]

megaparsec-6.0.0 (Mark Karpov [e-mail] @mrkkrp) is out of bounds for:

• [x] cue-sheet-0.1.1 (>=5.3 && < 6.0). Mark Karpov [e-mail] @mrkkrp. @mrkkrp. Used by: library
• [x] dotenv-0.3.4.0 (>=5.0 && < 6.0). Stack Builders [email protected] @stackbuilders. @jsantos17 @jsl @sestrella. Used by: library, executable, test-suite
• [ ] hledger-1.3 (>=5.0 && < 5.4). Simon Michael [e-mail] @simonmichael. @simonmichael. Used by: library
• [ ] hledger-lib-1.3 (>=5.0 && < 5.4). Simon Michael [e-mail] @simonmichael. @simonmichael. Used by: library, test-suite
• [x] hledger-web-1.3 (>=5.0 && < 5.4). Simon Michael [e-mail] @simonmichael. @simonmichael. Used by: library
• [x] hspec-megaparsec-0.3.1 (>=5.0 && < 6.0). Mark Karpov [e-mail] @mrkkrp. @mrkkrp. Used by: library, test-suite
• [x] inflections-0.3.0.0 (>=5.1 && < 6.0). Stack Builders [email protected] @stackbuilders. @jsantos17 @jsl @sestrella. Used by: library, test-suite
• [ ] isotope-0.5.0.1 (==5.*). Michael Thomas [e-mail] @Michaelt293. @Michaelt293. Used by: library, test-suite
• [x] language-puppet-1.3.8.1 (==5.*). Simon Marechal [e-mail] @bartavelle. @bartavelle. Used by: library, executable, test-suite
• [ ] language-thrift-0.10.0.0 (>=5.0 && < 6.0). Abhinav Gupta [e-mail] @abhinav. @abhinav. Used by: library, test-suite
• [x] stache-0.2.2 (>=5.0 && < 6.0). Mark Karpov [e-mail] @mrkkrp. @jsantos17 @jsl @sestrella. Used by: library, test-suite, benchmark
• [x] versions-3.1.1 (>=4 && < 6). Colin Woodbury [e-mail] @fosskers. @fosskers. Used by: library

[fosskers]

versions updated.

[DanBurton]

Entirely redundant with the above, but do also note:

hspec-megaparsec-1.0.0 (Mark Karpov [e-mail] @mrkkrp) is out of bounds for:

• [x] cue-sheet-0.1.1 (==0.3.*). Mark Karpov [e-mail] @mrkkrp. @mrkkrp. Used by: test-suite
• [x] dotenv-0.3.4.0 (>=0.2 && < 0.4). Stack Builders [email protected] @stackbuilders. @jsantos17 @jsl @sestrella. Used by: test-suite
• [x] inflections-0.3.0.0 (==0.3.*). Stack Builders [email protected] @stackbuilders. @jsantos17 @jsl @sestrella. Used by: test-suite
• [x] stache-0.2.2 (>=0.2 && < 0.4). Mark Karpov [e-mail] @mrkkrp. @jsantos17 @jsl @sestrella. Used by: test-suite

[bartavelle]

My parser seems to break when doing a naive move to v6. It will take me a while to find out why.

[simonmichael]

Some porting notes, perhaps better here than on reddit:

added where needed:

import Text.Megaparsec.Char

replaced Dec with MPErr and added:

-- build with old or new megaparsec
#if MIN_VERSION_megaparsec(6,0,0)
type MPErr = ErrorFancy Void
#else
type MPErr = Dec
#endif

added these as Text.Megaparsec.String/Text.Megaparsec.Text seem to have gone away:

type SimpleStringParser = Parsec MPErr String 
type SimpleTextParser = Parsec MPErr Text

[fosskers]

The library itself recommends setting the error type (where MPErr is used above) to Void if its not needed.

[simonmichael]

Here's a Text.Megaparsec.Compat module, useful for supporting megaparsec <6 and >=6.

[mrkkrp]

@bartavelle If you have any trouble switching please open an issue, I'll help.

[DanBurton]

@fosskers: this was unfortunate, but I removed "versions" from the build plan since it could not satisfy both the new version of base (which we are upgrading to now) and the old version of Megaparsec (which we are upgrading later).

[bartavelle]

@mrkkrp I'll try to pinpoint the problem next week. The only meaningful change I did was changing an integer into a decimal, but this gave really weird results (change in whitespace handling, 0x800 parsed as 0). I am sure I did something wrong, so will do that when I have plenty of time to do it properly.

[fosskers]

@DanBurton but I bumped my base upper bound too... :(

Hmm... actually I don't think I can support any megaparsecs less than 6 now, since a user requested that I expose some of the internal parsers that versions uses in case they wanted to include version number parsers in a larger parser of their own. Exposing these puts a bit of a stranglehold on versions' major version and dep ranges.

[mrkkrp]

Updated:

• hspec-megaparsec
• stache
• cue-sheet

[bartavelle]

@mrkkrp Turned out my parser had bugs, but they did not surface with megaparsec 5!

[bartavelle]

released language-puppet-1.3.9

[juanpaucar]

Fixed for dotenv in version 0.4.0.0

[CristhianMotoche]

inflections-hs updated! inflections-0.4.0.0 in hackage

[mrkkrp]

Can we perhaps update the checkboxes in the first post?

[mihaimaruseac]

Checkboxes updated

[CristhianMotoche]

@mihaimaruseac inflections and dotenv-hs are already updated. Could you please give check our packages in the todo? or are we missing something else?

[mihaimaruseac]

I see they are updated already.

[CristhianMotoche]

Yeah. I see they are already checked. Thanks! @mihaimaruseac

[PierreR]

I would love to see megaparsec updated to 6.1.1 in stackage-nightly. It looks like many packages are currently on hold for nightly waiting for the others to make the megaparsec update. As time goes on, this seems less fair ;-)

[simonmichael]

You are right! and I am sorry for being the laggard here. I updated the hledger packages early but have been hung up on the release process. I will release them today.

[cdornan]

@simonmichael totally cool: will get them right back as soon as the releases have been made. Thanks!

[simonmichael]

Thanks @cdornan.

[cdornan]

Megaparsec-6 has been blocked for 4 weeks so we have decided to (temporarily we hope) evict the remaining blockers and restore them as soon as they are compatible with megaparsec-6.

I have raised issue in the three projects concerned:

• hledger (though see above);
• isotope
• language-thrift

[mrkkrp]

Now that Megaparsec 6 is in nightly builds, can we allow newest stache (1.1.1)?