Release 0.2.0.0 / new maintainer wanted about resolv

[lyokha]

@andreasabel I propose to go ahead as Herbert keeps silent. Further steps would be:

1. Merge #16, #12 (perhaps others, I cannot do this as I don't have permissions in this repository)
2. I would also consider merging #17 by @dminuoso after fixing a small issue (throw -> throwIO)
3. Bump upper base constraint to 5 in Cabal file (and those of other packages too if needed)
4. Build and make sdist and haddocks (I can do this in my branch after pulling from upstream or in a cloned upstream)
5. Upload built sdist (perhaps, haddocks too) on Hackage (I can do this as I have permission).

[lyokha]

I mailed Herbert.

[lyokha]

Good news! Herbert will look at this soon, probably, this week.

[lyokha]

@andreasabel, yesterday I got maintenance permissions to resolv at Hackage, but I am waiting Herbert as he wanted to have a look at the issues.

[lyokha]

@ulysses4ever, done, now base >= 4.5 && <5

[andreasabel]

@lyokha : I invited you to this repository with write permission, so you can merge PRs now.
You should now also be able to comment on PRs (after accepting the invitation, of course).

I think it is best to keep the code for resolv here on haskell-hvr which is meant as a collaborative organization by Herbert (where we work together on maintaining packages).

If you make a candidate upload to hackage first and ping me, I can have a look at the candidate before you finally publish.
Usually, haddocks are built on Hackage for the final release, so it is better to not upload them manually there (unless build on Hackage fails).
However, for candidates one has to upload the documentation manually.

[lyokha]

Ok, in a 1 or 2 hour when i reach to my laptop, I would also wait #17 with improvements requested, when @dminuoso looks at it

[lyokha]

@andreasabel I added you

[andreasabel]

@andreasabel Should I register at Stackage for this?

It is enough to make a PR against https://github.com/commercialhaskell/stackage/ with changes to the build-constraints.yaml file.

[lyokha]

This one is good: https://hackage.haskell.org/package/resolv-0.2.0.1/candidate/src/configure: it contains lines with netdb

ac_fn_check_decl "$LINENO" "h_errno" "ac_cv_have_decl_h_errno" "
#include <netdb.h>

" "$ac_c_undeclared_builtin_options" "CFLAGS"

[andreasabel]

I added release instructions:

resolv/DEVELOPER.md

Lines 4 to 7 in 7c26271

	Before uploading to hackage, make sure that `configure` script is up to date, by running:
	```
	autoreconf -i
	```

[lyokha]

Hurray!

$ ghci
GHCi, version 9.6.1: https://www.haskell.org/ghc/  :? for help
Loaded GHCi configuration from /home/lyokha/.ghci
Prelude> import Network.DNS 
Prelude Network.DNS> :set -XOverloadedStrings 
Prelude Network.DNS> queryA $ Name "www.ayyayagoogle.com"
*** Exception: DnsHostNotFound
Prelude Network.DNS> 

Memory leaks have gone as well.

[phadej]

I removed myself from resolv maintainer group: https://hackage.haskell.org/package/resolv/maintainers/

Either ask herbert to add you there, or takeover the package.

[andreasabel]

I removed myself from resolv maintainer group: hackage.haskell.org/package/resolv/maintainers

Not sure why this was necessary. This needlessly complicates things. Sigh.

Either ask herbert to add you there, or takeover the package.

You could as co-maintainer just have added me. Herbert has gone into hiding. Very funny!

[phadej]

You could as co-maintainer just have added me

I was helping Herbert, I didn't ever "took over" the maintainership. I don't want to add more maintainers transitively as I weren't explicitly allowed to do so, and particularly because we do disagree on how stuff should be maintained.

This is easier to me, I feel bad for making things harder, but OTOH I would feel worse by doing something I'm not 100% happy with.

[andreasabel]

Since no shortcut is possible anymore, I suggest someone with domain knowledge takes over.

Alexey @lyokha, if you are interested in taking over this package, I'd second this.
The process is described here: https://wiki.haskell.org/Taking_over_a_package

[lyokha]

@andreasabel, I do not mind. So, should I make further steps, e.g. mailing to Herbert, or will you do it yourself?

[andreasabel]

@lyokha wrote:

@andreasabel, I do not mind. So, should I make further steps, e.g. mailing to Herbert, or will you do it yourself?

Thanks, great!
Yes please email Herbert, and follow the steps of the package takeover.

[lyokha]

@andreasabel , news at this moment is short: I sent mails to @hvr January 30 and then, 3 weeks later, February 20, to [email protected]. Nobody responded so far.

[ulysses4ever]

@lyokha could you in the meantime, please, bump base upper bound on the current Hackage release via a revision? That would unstuck some things...

[lyokha]

3a. Update Changelog.md.

[andreasabel]

@lyokha: Would you add me as another maintainer on hackage so that I could act as co- or fall-back maintainer? That would also enable me to upload candidate releases for testing purposes.
https://hackage.haskell.org/package/resolv/maintainers/

[andreasabel]

@lyokha : Do you want to take over the stackage maintainership? https://github.com/commercialhaskell/stackage/blob/99c752318f259be25cf13c1d994a50f2f0583829/MAINTAINERS.md#adding-a-package

Since resolv is already part of Stackage, it amounts to putting the following entry under a section with your name and email:
https://github.com/commercialhaskell/stackage/blob/99c752318f259be25cf13c1d994a50f2f0583829/build-constraints.yaml#L5342

You will then be notified on stackage issues around resolv.

[lyokha]

@andreasabel Should I register at Stackage for this?

[lyokha]

@andreasabel it looks that the #17 is not present in the build. See https://hackage.haskell.org/package/resolv-0.2.0.0/reports/1. Probably, we need somehow trigger autoreconf at hackage builder for this.

[lyokha]

@andreasabel , it seems that it is not a Hackage issue, perhaps you forgot to autoreconf -i before making dist and the configure shipped with the package did not test for netdb and h_errno. In the log, line checking for struct __res_state.res_h_errno... yes is absent. Would you re-upload the fix in 0.2.0.1?

[andreasabel]

Oops, wasn't aware of the autoreconf step. This is easy to blunder (configure isn't version-controlled, and cabal doesn't automatically run autoreconf either).
Candidate: https://hackage.haskell.org/package/resolv-0.2.0.1/candidate

[andreasabel]

Published: https://hackage.haskell.org/package/resolv-0.2.0.1

[andreasabel]

it looks that the #17 is not present in the build.

It would be good to have a test that ensures such things. I do cabal test before uploading, this could catch oversights...

[andreasabel]

I deprecated 0.2.0.0 on hackage to indicate that this is not a good release.

[lyokha]

Oops, wasn't aware of the autoreconf step. This is easy to blunder (configure isn't version-controlled, and cabal doesn't automatically run autoreconf either). Candidate: https://hackage.haskell.org/package/resolv-0.2.0.1/candidate

Yes, I did not keep in mind this as well, just guessed what happened.

[lyokha]

I deprecated 0.2.0.0 on hackage to indicate that this is not a good release.

Strictly speaking, 0.2.0.0 wasn't bad, it just did not provide what it claimed in the changelog.

[andreasabel]

Congrats, this issue is resolved to full satisfaction!