Stefan Hagen wrote:
> Solene Rapenne wrote:
> > On Thu, 16 Sep 2021 19:33:07 +0100
> > Stuart Henderson <s...@spacehopper.org>:
> > 
> > > On 2021/09/16 19:02, Omar Polo wrote:
> > > > 
> > > > Solene Rapenne <sol...@perso.pw> writes:
> > > >   
> > > > > On Thu, 16 Sep 2021 09:04:37 +0200
> > > > > Omar Polo <o...@omarpolo.com>:
> > > > >  
> > > > >> Solene Rapenne <sol...@perso.pw> writes:
> > > > >>   
> > > > >> > Import a simple python dependency for editors/apostrophe
> > > > >> >
> > > > >> > I generated it with portgen, checked version, DESCR, PLIST, 
> > > > >> > licensing.
> > > > >> > I didn't add pandoc as a dependency because it's not required to
> > > > >> > have it at build stage, and I'm not sure it would be fine to add
> > > > >> > pandoc as a run dep either.    
> > > > >> 
> > > > >> without text/pandoc as TEST_DEPENDS the tests fails; well, they fails
> > > > >> nevertheless because it tries to fetch stuff from github.  No idea 
> > > > >> how
> > > > >> to fix it, sorry.
> > > > >> 
> > > > >> Anyway, I'd argue that it needs text/pandoc as {RUN,TEST}_DEPENDS and
> > > > >> not RDEP on it from apostrophe since it searches for the binary and
> > > > >> throw an error if not found.  Attaching another tarball with pandoc
> > > > >> added as deps.
> > > > >> 
> > > > >> Cheers,
> > > > >>   
> > > > >
> > > > > I added TEST_DEPENDS , it can make 23 checks before failing
> > > > > when trying to download its own sources to use the README.md file...
> > > > > but I suppose 23 checks working is better than nothing.  
> > > 
> > > No need to list RUN_DEPENDS in TEST_DEPENDS
> > > 
> > > > I was distracted by the error output and the /writes_to_HOME lines.  It
> > > > seems that all but that test are succeeding, which is good I guess.  
> > > 
> > > PORTHOME=${WRKDIR} probably helps with the /writes_to_HOME
> > > 
> > > > > It's now depending on pandoc for RUN_DEPENDS  
> > > > 
> > > > I've seen these two nits only now, but I'm not familiar with the python
> > > > infrastructure in ports so they may be wrong/opinable:
> > > > 
> > > >  - comment starts with uppercase; also maybe specify that's a python
> > > >    wrapper?  
> > > 
> > > Python is a proper noun so uppercase is preferable there
> > > 
> > > >  - use ``py-wheel${MODPY_FLAVOR}'' instead of ``py-wheel,python3''  
> > > 
> > > correct
> > > 
> > > > If by any chance I'm not wrong, here's an updated tarball.
> > > > 
> > > > Cheers!
> > > >   
> > > 
> > 
> > new tarballs attached
> > 
> > - py-pypandoc comment has now uppercased Python, added PORTHOME
> > - removed a trailing whitespace in apostrophe
> > 
> > Apostrophe complains about missing texlive for exporting PDF but it
> > does so with a nice GUI explaining what is missing. I prefer not add
> > texlive as a run dependency, ok?
> 
> For apostrophe:
> 
> You're getting it from github, which is a mirror. The primary source 
> is on gitlab. Would it make sense to set the gitlab URL as homepage? 
> This is where issues etc. are hosted.
> 
> Nitpick: You could reduce the indentation by one level for all 
> entries.
> 
> For py-pypandoc
> 
> attached again with the following changes:
> - Added test dependency devel/py-pip.
> - Patched the failing test case, so it runs with a local file:// url.
>   The README.md is already present, and the file:// url should run 
>   the same code path within py-pypandoc.

The testrun with apostrophe on an empty system was successful.

OK sdk@ with the tweaks from my last mail

Attachment: py-pypandoc.tgz
Description: application/tar-gz

Reply via email to