Created attachment 162066 [details] Proposed patch (since 394382 revision) Patch to update deskutils/treesheets port from 0.0.20150711 to 0.0.20151008 version. Look following link for changes: https://github.com/aardappel/treesheets/compare/3a6be83...10fbc77
Created attachment 162067 [details] The poudriere testport log (FreeBSD 10.2 amd64)
Comment on attachment 162066 [details] Proposed patch (since 394382 revision) There were changes after ports r399346, but the patch is same.
A a side note, you do not have to attach poudriere testport logs to the PR, stating "builds fine on 9/10" is enough.
A commit references this bug: Author: mat Date: Fri Oct 16 14:19:13 UTC 2015 New revision: 399479 URL: https://svnweb.freebsd.org/changeset/ports/399479 Log: Update to 0.0.20151008. PR: 203787 Submitted by: maintainer Sponsored by: Absolight Changes: head/deskutils/treesheets/Makefile head/deskutils/treesheets/distinfo
Hello, Mathieu Arnold. (In reply to comment #3) > you do not have to attach poudriere testport logs to the PR, > stating "builds fine on 9/10" is enough. I understand, that your proposed approach works, but I just use a poudriere testport log(s) as a proof of my work at the time of creation and (almost) latest ports changes. Usually, I test on the OS version, which I use (e.g. 10.2, currently). In case, there are errors for other supported OS version (e.g. previous 9.3), based on results of http://portsmon.freebsd.org, then I try to create a poudriere "jail" for it to test. This way, I may act in case of real errors (e.g. spotted by me, reported by portsmon or other reviewer), without the need to create a proof each time, when something changes in ports framework (or other OS version). In other hand, creation of poudriere testport logs for many supported OS versions is time consuming process (especially, for the port(s), which may use WebKit dependencies), therefore, some proof is better, than nothing, I guess. Thanks for commit.
Well, it's not kindergarden, we don't require proof that you tested :-) And, in the end, there is nothing interesting in a successful log, it's just many many kilobytes of text basically saying "ok". I did not say you should not test, just that the successful build logs are not needed.
I guess, this is off-topic, but anyway :) (In reply to comment #6) I think, the opinion about poudriere testport logs depends from the reviewer(s). I don't know who will review the PR from the start, therefore I decide on my own. In my experience, I had requests for plain text logs, which is easy to see in a browser. In the end, I decided, that I need to post uncompressed logs, when the size is less than 100 Kb and compressed otherwise. Now, you suggested to not post successful logs at all. This is great, but in case they might be needed (e.g. to change the status of the PR), this might require more work on my side, related to synchronization of latest ports changes and possible issues with it (e.g. some other dependent port stopped working after synchronization). Currently, I still see requests for portlint results and/or poudriere testport logs (e.g. bug 203816, comment 1) and people who provide them (e.g. bug 200488, comment 4). Therefore, I just post these logs, when they available. May be, there is no requirement to post such logs for a maintainer. I don't know. Thanks for your conversation and opinion.
There is no requirement. And it is not an opinion, it is a fact, I know, because I am the member of portmgr writing the documentation :-)
Ok, I understood. I found following paragraph in the current FreeBSD Porter's Handbook (9.5. Poudriere): > Ports committers sometimes ask for a Poudriere log alongside a patch > submission to assess whether the patch is ready for integration into > the ports tree So, instead of adding poudriere testport log(s) for each PR, I may write about its availability on request (but not after PR completion, of course). Still, I can't guarantee the prompt execution of such request after long period of time, if no attention to the PR from committer(s). In this case, there is a need to test the patch on their own. I already see the consequences of such approach and where my current approach in much better position, in regard to faster execution. Let's see how it goes on the next available patch(es) for port(s), which I maintain.
(In reply to lightside from comment #9) > Ok, I understood. > > I found following paragraph in the current FreeBSD Porter's Handbook (9.5. > Poudriere): > > Ports committers sometimes ask for a Poudriere log alongside a patch > > submission to assess whether the patch is ready for integration into > > the ports tree Damn, I thought I had removed that bit, I will rewrite that part today. > So, instead of adding poudriere testport log(s) for each PR, I may write > about its availability on request (but not after PR completion, of course). > Still, I can't guarantee the prompt execution of such request after long > period of time, if no attention to the PR from committer(s). In this case, > there is a need to test the patch on their own. I already see the > consequences of such approach and where my current approach in much better > position, in regard to faster execution. Let's see how it goes on the next > available patch(es) for port(s), which I maintain. The thing is, the committer *will* test that it builds, with poudriere, whether it is on the same day or six months afterwards. There is no *value* in a successful build log, the one attached in this PR can be replaced with "builds fine in poudriere on 10.2 amd64" it has the same value.