Ping

On 9/7/23 22:05, Igor Zornik wrote:
On 9/6/23 18:14, Omar Polo wrote:
 Hello,
On 2023/08/19 10:04:54 +0200, Igor Zornik <[email protected]> wrote:
 Greetings ports@,

 Long time listener, but first time caller so I hope I didn't get to many
 things wrong.
Sorry for the delay. I wanted to review this the day it was sent
 but only managed to look at it today.
 Attached I have a port of alps, an e-mail web client. It's not as
 feature-full as some other web clients, but I like this one because it
 has not run-time dependencies which makes it easier to deploy and maintain.

 Notes:
 - Port initially created with `portgen`. When modified I relied heavily
 on gitea port.
 - Upstream isn't very responsive to my change proposals so I provided
 some fixes and features locally.
 - There isn't any special functionality associated with the service
 account so I just made it to run as `www`. Should I still run it under a
 dedicated service account?
 - I had to do some `sed` trick in the `rc` script to get the `rc`
 command to pick up the daemon properly. Could anyone confirm if i have
 taken the right path here?
 - Tested on amd64 using -current and 7.2.

 Any comments or suggestions?
a few nits, but otherwise it's a solid submission. - I'd add some commentary to the patches. Just something brief like
     "adding FastCGI support", so that it's clear what it does when
     skimming the files.
- I'd drop the patch for go.mod; I'm not sure go.port.mk even looks
     at the fetched go.mod file, let alone parsing it once patched.
- there's a small error in the fastcgi patch: + addr = strings.TrimPrefix(addr, "fcgi+unix:/")
     should be
 +               addr = strings.TrimPrefix(addr, "fcgi+unix://")
i.e. remove two slashes. In practice it's not an issue. - While I really like the fastcgi support (I'm a bit lazy and never
     bothered to learn how to use relayd), I don't think we've usually
     patched the ports to add features like these.  Have you sent the
     patch upstream?  I can't find it searching for 'fastcgi' on
     sourcehut.
To be fair I don't think it's an issue, but still felt like
     pointing this out.
- It doesn't work without altering the flags to specify the upstream
     server.  This has confused me initially, so add that info to the
     README.
I've also tweaked the fastcgi example to use `rcctl set alps flags'
     and add a dummy example.com at the end to minimize the chance of
     getting things wrong.
- (opinable) I've tweaked the http example to listen on port 80.
     Otherwise it's not copy-paste-able without further fiddling.
- I'm not sure about ${SYSCONFDIR}/httpd.conf. Even if portcheck
     complains, the correct path is /etc/httpd.conf: it is hardcoded in
     the httpd(8) binary.
- make update-plist changed @cwd ${LOCALSTATEDIR}/www into /var/www
     and I'm not sure how much we care about using the variables instead
     of the hardcoded paths.  Left as /var for the moment.
- no opinion on the sed machinery to escape '+'s in the pexp. Seems
     to work :-)
- changed the version to 0.0.0.<date> so if upstream starts to do
     proper releases we don't have to use EPOCH.
I'm reattaching a tarball with these points addressed. It's ok op@ to
 import if someone wants to, otherwise i'll be happy to import it
 myself (needs an ok to do so).
I don't use web ui for mails if I can, but this is one of the nicer
 I've seen.

Ciao!

Thanks for taking your time with the port. I'm just replying to say that even with the changes it still works on my machine. Also to clarify a few points.

 > I'd add some commentary to the patches.

Nice tip. I didn't know you could include comments like that in a patch. It does mess up syntax highlighting in some editors though.

 > Have you sent the patch upstream?

Patch for FCGI has not been sent to upstream yet because I'm still waiting for the response on the first patch proposal (the `go.mod` one). In due time I'll probably just consider it rejected and move on to the next proposal.

> It doesn't work without altering the flags to specify the upstream server. This has confused me initially, so add that info to the README.

I've only included things that aren't mentioned in the original documentation. But if a little coping saves some confusion then it's OK. I would however like to suggest some changes for the new README for consistency sake (see attached diff):

Anyhow, looking forward to submitting additional ports.

With regards,


Reply via email to