On Wed, Dec 9, 2020 at 5:33 PM sshuck <stevenjsh...@gmail.com> wrote:
> Fantastic, thanks for the quick reply, digging in, and the suggestions. > > Just to get over the hump, I did two more releases, the first manually > relaxing the text lower bound (failed), and the second removing > --pvp-bounds altogether (succeeded). In the past I've been advised by > Hackage curators to set bounds for every dependency for a package I took > over maintainership of. Which in principle I agree with but in practice > it's tedious and also hard to know what the actual requirements are. To > illustrate, the candidates upload feature does not do CI, so I can't know > if it'll fail, and I can't delete non-building releases when they do fail. > > My outstanding questions are: > 1. Where is there documented a sensible policy for setting and maintaining > bounds that will satisfy mainstream tools and personnel; > This is unfortunately more complicated than it should be. The short version of the backstory is: some people want complete compliance with PVP rules. Others think the rules are useless, and don't have a reason to implement them. Personally, I fall almost entirely in the second camp. I tried to find some compromise positions (including the --pvp-bounds flag), but nothing short of full compliance satisfies the PVP side. Ultimately I gave up, and now I just ignore PVP rules. > 2. Do any tools exist that can help automate this; > I don't think anything beyond --pvp-bounds. > 3. How can I have these bounds committed to source control versus manually > tweaking the tarball before upload, and hopefully still use Stack in my > project; > You're free to add the bounds to your cabal file or package.yaml file if you like. Stack will respect the bounds if present, and give you a bounds error if you violate them. Whether it's worth that effort is up to you. I've decided for myself that it's not worth it. > 4. How can I test Hackage's CI locally to know if my upload will build > there? > > Not in any way I'm aware of. > Thanks again for the assistance. > > On Wednesday, December 9, 2020 at 1:22:31 AM UTC-5 mic...@snoyman.com > wrote: > >> You can see the cabal build error by clicking on the "build log" link on >> your Hackage page, which goes to >> https://hackage.haskell.org/package/pcre2-1.0.0/reports/2. Here's the >> juicy bit: >> >> [__1] rejecting: text-1.2.3.2/installed-1.2.3.2 (conflict: pcre2 => >> text>=1.2.4.0 && <1.3) >> [__1] trying: text-1.2.4.0 >> [__2] next goal: base (dependency of pcre2) >> [__2] rejecting: base-4.14.1.0/installed-4.14.1.0 (conflict: text => >> ghc-prim>=0.2 && <0.6, base => ghc-prim==0.6.1/installed-0.6.1) >> >> The PVP bounds added by Stack are putting a lower bound on text of >> 1.2.4.0. That version of text has an upper bound on ghc-prim of < 0.6. GHC >> 8.10.2 ships with ghc-prim 0.6. And the Haddock build server only builds >> with GHC 8.10.2. Your options to fix this are: >> >> * Convince the text package maintainer to relax the upper bounds on >> ghc-prim so it will build with GHC 8.10.2 >> * Stop using the --pvp-bounds flag so you don't get the lower bound on >> text >> >> I no longer recommend the --pvp-bounds flag, specifically because of >> causing problems like this. >> >> On Wed, Dec 9, 2020 at 8:16 AM sshuck <steven...@gmail.com> wrote: >> >>> Hello fine haskell-stack people, >>> >>> I am trying to upload my Stack-based library to Hackage, but it's not >>> able to resolve dependencies and thus doesn't build documentation. >>> Here's some info: >>> >>> 1. Latest Stackage LTS resolver (16.25) >>> 2. Cabal file is completely auto-generated >>> 3. `stack sdist --pvp-bounds both --test-tarball` executes and tests >>> on my computer just fine, both Windows 10 and WSL2 (Ubuntu 20.04) >>> 4. Uploaded the resulting tarball from .stack-work/ manually via >>> Hackage web UI >>> >>> Here's the link to the package page, containing further links to Github, >>> the auto-generated PVP ranges, and the build failure logs: >>> https://hackage.haskell.org/package/pcre2 >>> >>> Any tips on what I'm supposed to do to solve this would be greatly >>> appreciated. >>> Thanks, >>> Shlomo >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "haskell-stack" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to haskell-stac...@googlegroups.com. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/haskell-stack/150cecbe-d4c6-409f-b98a-9b9aa7fe9b09n%40googlegroups.com >>> <https://groups.google.com/d/msgid/haskell-stack/150cecbe-d4c6-409f-b98a-9b9aa7fe9b09n%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> -- > You received this message because you are subscribed to the Google Groups > "haskell-stack" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to haskell-stack+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/haskell-stack/c464ccd6-42bb-4491-89b8-855c45a6b974n%40googlegroups.com > <https://groups.google.com/d/msgid/haskell-stack/c464ccd6-42bb-4491-89b8-855c45a6b974n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "haskell-stack" group. To unsubscribe from this group and stop receiving emails from it, send an email to haskell-stack+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/haskell-stack/CAKA2Jg%2ByfBfw7-5PpDbVHfvZ3QF3GYhBWF_eKSy%3DGKsAgM8aMg%40mail.gmail.com.
