[Pkg-haskell-maintainers] Bug#753299: libghc-highlighting-kate-dev: Highlighting Ocaml fails

[John MacFarlane]

OK, I have the solution!  The \o{377} format for specifying
octal escapes is only supported starting in PCRE 8.34, and debian
has an older version.  \377 can still be used.  So I can adjust
highlighting-kate code to use e.g. \377, as long as we don't have
characters greater than octal 377.

+++ John MacFarlane [Jul 19 14 08:17 ]:
>Another solution would be to modify the highlighting-kate code to
>avoid using character classes with octal escapes.  (I could simulate
>them by just listing all the characters.)  Perhaps that is the best
>short-term solution.
>
>+++ John MacFarlane [Jul 19 14 08:14 ]:
>>Update:  It seems that pcre-light is no longer being actively maintained
>>upstream.  pcre-regex-builtin is not in Debian.  I thought the solution
>>might be to add an option to highlighting-kate to use the regex-pcre library
>>(in debian as libghc-regex-pcre-dev).  However, this seems to have the same
>>problem:
>>
>>Prelude Text.Regex.PCRE.ByteString Data.ByteString.UTF8> Left e <- compile (compUTF8 + compAnchored) 0 (fromString "[\\o{0370}-\\o{0377}")
>>Prelude Text.Regex.PCRE.ByteString Data.ByteString.UTF8> e
>>(11,"range out of order in character class")
>>
>>Is the problem perhaps that the underlying C library pcre has not been
>>compiled with UTF8 support?
>>
>>+++ John MacFarlane [Jul 18 14 16:53 ]:
>>>I'd tested before with pcre-regex-builtin, which worked.
>>>I recompiled with pcre-light, and got the failure you report.
>>>
>>>The problem can be exhibited in pcre-light:
>>>
>>>Prelude Text.Regex.PCRE.Light Data.ByteString.UTF8> match (compile
>>>(fromString
>>>"[\\o{0370}-\\o{0377}") [utf8,anchored]) (fromString "hi") []
>>>*** Exception: Text.Regex.PCRE.Light: Error in regex: range out of order
>>>in
>>>character class
>>>
>>>The unicode ranges seem to be confusing it, even with 'utf8' set.
>>>
>>>Submitted an issue upstream:
>>>https://github.com/Daniel-Diaz/pcre-light/issues/1
>>>
>>>In the mean time, I suggest that debian packagers use pcre-regex-builtin
>>>instead (though perhaps there was a reason for using pcre-light?).
>>>
>>>+++ Samuel Hym [Jul 01 14 18:06 ]:
>>>>Hi John,
>>>>
>>>>>Are you certain that the version you are using in GHCI is 0.5.8.2,
>>>>>and not an older version (perhaps from your user database)?
>>>>
>>>>I rechecked, using the same version of the package but on a 
>>>>different machine, different arch, just in case. But still with 
>>>>debian-built packages, and with the same result.
>>>>
>>>>Here is what ghci has to say, with the full versions of the modules used:
>>>>
>>>>GHCi, version 7.6.3: http://www.haskell.org/ghc/  :? for help
>>>>Loading package ghc-prim ... linking ... done.
>>>>Loading package integer-gmp ... linking ... done.
>>>>Loading package base ... linking ... done.
>>>>Prelude> :m +Text.Highlighting.Kate
>>>>Prelude Text.Highlighting.Kate> highlightAs "ocaml" "a"
>>>>Loading package array-0.4.0.1 ... linking ... done.
>>>>Loading package deepseq-1.3.0.1 ... linking ... done.
>>>>Loading package containers-0.5.0.0 ... linking ... done.
>>>>Loading package bytestring-0.10.0.2 ... linking ... done.
>>>>Loading package transformers-0.3.0.0 ... linking ... done.
>>>>Loading package mtl-2.1.2 ... linking ... done.
>>>>Loading package text-0.11.3.1 ... linking ... done.
>>>>Loading package parsec-3.1.3 ... linking ... done.
>>>>Loading package blaze-builder-0.3.3.2 ... linking ... done.
>>>>Loading package blaze-markup-0.5.1.6 ... linking ... done.
>>>>Loading package blaze-html-0.6.1.2 ... linking ... done.
>>>>Loading package pcre-light-0.4 ... linking ... done.
>>>>Loading package utf8-string-0.3.7 ... linking ... done.
>>>>Loading package highlighting-kate-0.5.8.2 ... linking ... done.
>>>>[*** Exception: Text.Regex.PCRE.Light: Error in regex: range 
>>>>out of order in character class
>>>>
>>>>You think something else might be to blame in the module stack?
>>>>Do you know if there is a way to ask pcre-light to be somewhat 
>>>>more informative about the regex producing the issue?
>>>>
>>>>By the way, many thanks for pandoc!
>>>>
>>>>Best regards
>>>>Samuel