F/remove msg error etc null sentinel

[bkil-syslogng]

• msg_error("description", evt_tag_str("when", "before"), NULL);
• msg_error("description", evt_tag_str("when", "after"));

Contains a sed moster and some vararg magic.

[MrAnno]

👍

[bkil-syslogng]

Rebased to master

[bazsi]

I like the change in general, even though I didn't have too much time to
check how it works, and neither did I review the individual call sites.

Bazsi

On Wed, Mar 2, 2016 at 1:59 PM, Tamas Nagy notifications@github.com wrote:

Rebased to master

—
Reply to this email directly or view it on GitHub
#941 (comment).

[bkil-syslogng]

Rebased to master.

Fixed conflicts by dropping the last commit and reapplying the script described in the commit message.

[ihrwein]

👍 (Could you rebase it again? Thanks!)

[bkil-syslogng]

@ihrwein Rebased to master. No conflict or omission detected when running the transformation this time.

[bazsi]

I am afraid I've just merged code that will not be adjusted for this. I would merge this again and fix up the fallout. The change is trivial, makes sense it's just scattered around.

Do you have a script to run on the "http" destination?

[bkil-syslogng]

@bazsi You can re-run the commit message as many times as you'd like. Do take care to remove any extra space if your tools show messages that way. Actually I wrapped a long sed line in the middle, that's the (only) one you need to consider.

I did give this a second thought few days ago, and we may consider saving this under scripts/ and eventually connect it with our Pull Request examination suite in a way that a bot would notify if you used an extra ,NULL. It's mostly a matter of whether the git diff after the transformation is empty or not. In the long run, we could probably connect a bunch of custom LLVM analysis to catch more important typos/common errors as well.

[bkil-syslogng]

@bazsi I ran it on the HTTP module. You can check and merge the result and the script file in #1021.

[bazsi]

I don't think we need to add this validation from this point on. We don't need it from now period, it may happen that some as-of-now out-of-tree code contains the extra NULL, but we won't write new ones.

[bazsi]

thanks a lot and sorry for breaking things. probably not a good idea to
keep PRs out-of-tree for weeks as the chance of stuff breaking when merging
them in improper order is a lot larger.

Bazsi

On Wed, Apr 13, 2016 at 4:23 PM, Tamas Nagy notifications@github.com
wrote:

@bazsi https://github.com/bazsi I ran it on the HTTP module. You can
check and merge the result and the script file in #1021
#1021.

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#941 (comment)