Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cross-project-issues-dev] 1000 line limit for contributions

I don't think reviews should ever be split up so that one review contributes functionality and another contributes tests. If Gerrit is going to run a build on a review, the build should test the new functionality added by the review. In my experience, the IP review process is usually very fast so the benefit of having both the functionality and the tests in a single change, which can be cherry-picked to other branches, reverted, etc. far outweighs the slight delay imposed by the IP review. I do encourage contributors to split large reviews into smaller chunks just to make them easier to review, but each chunk should be a cohesive piece of functionality including tests.

Sam


--
Sam Davis
Software Engineer, Tasktop Dev
Committer, Eclipse Mylyn
http://tasktop.com

On Thu, Nov 19, 2015 at 7:17 AM, Doug Schaefer <dschaefer@xxxxxxx> wrote:
We need to keep in mind the spirit of the rule which is as Pascal mentioned. I always thought the lines of code was a guideline and certainly doesn’t excuse the committer from taking a good look at the code and even if it’s less than 1000 lines, if it looks like a algorithm that could be taken from somewhere else that it gets reviewed.

From: <cross-project-issues-dev-bounces@xxxxxxxxxxx> on behalf of Fred Bricon <fbricon@xxxxxxxxx>
Reply-To: Cross project issues <cross-project-issues-dev@xxxxxxxxxxx>
Date: Thursday, November 19, 2015 at 10:12 AM
To: Cross project issues <cross-project-issues-dev@xxxxxxxxxxx>
Subject: Re: [cross-project-issues-dev] 1000 line limit for contributions

I'm not sure how the workload from the IP team is split, between running automated IP checks (blackduck or similar) and manual work. But I think it'd be interesting if at least the automated part could be included as part of a CI build checking gerrit contributions. That'd would surely offload the IP team work significantly.

On Thu, Nov 19, 2015 at 10:01 AM, Pascal Rapicault <pascal@xxxxxxxxxxxx> wrote:
It is my understanding that the 1000 lines limit is here to prevent code to be copied from an external place and be added to Eclipse.  This is a good measure.
However, I think there is a number of contributions where the code has obviously not been copied. For example, this is the case with the contribution Jan mentioned. There is no way the code could be coming from somewhere else. It is way too specific to Tycho and the feature being added.

IMO, the IP process could be relaxed and thus the IP teamwork load reduced, if committers were trusted on what to put through the IP process and what not.

With such a rule, a committer who has doubts about 200 lines of code could take it through the IP process, and yet for things that are obviously "new" code, would not. IMO, such a process based on committer trust could help focus the efforts of the IP team on things that are potentially problematic.




On 15-11-19 05:47 AM, Ed Merks wrote:
Yes, I believe it's an important aspect of Eclipse that makes it stand out as the best place to be if you want the broadest possible community of adopters.   Of course this benefit doesn't come without a cost and of course that can be frustrating.   In a specific case of a contribution that consists of a relatively smaller changes to the framework/tool with a relatively larger addition of test case(s), it would seem reasonable to split the two, if it's important that the change to the tools/framework show up as quickly as possible.

I certainly don't suggest gaming the system, though I do tend to point out to the IP committee all the ways it can be gamed, and will be gamed by developers who are frustrated and don't take the issue seriously.  I ask questions such as how long can a line be? One can fit quite a lot on a line line and reformat it later. Also, why should a blank line count for anything?  Is a line with just a curly brace on it really IP?   And yes, of course I make them aware that contributions can be split into smaller chunks...

Perhaps this specific review period overlapped with EclispeCon Europe where we had the pleasure of spending personal time with the with the IP staff...


On 19/11/2015 11:31 AM, Christian Campo wrote:
Wouldnt it be worth to hear what the IP Team has to say why this took so
long ? I see that Sharon appologized on the CQ that it took so long. That
made me believe that this was an exception.

Does every CQ with 1000 lines take so long ? What is the experience of
others about reviews with code contributions.
As I remember vaguely (and that might be incorrect) the IP team runs
automatic scans over the code, but I am not sure what else they do.

I for once believe the work of the IP Team is important and one of the
core values of the EF vs say Github and I take it serious.

Just my 2 cents

christian

Am 19.11.15, 11:22 schrieb "cross-project-issues-dev-bounces@xxxxxxxxxxx
on behalf of Sievers, Jan" unter
<cross-project-issues-dev-bounces@xxxxxxxxxxx on behalf of
jan.sievers@xxxxxxx>:

If everybody tells me there are ways to dodge around that rule (and of
course I know there are), the question arises why do we have the rule in
the first place. Seems a little absurd to me.

the effort is not minimal if I have to artificially split up commits.
Or maybe you expect me to explain to contributors:

"look, we have this process but nobody takes it serious anyway. so please
split up your commit into several < 1000 LOC chunks" ?

Best Regards,
Jan



On 19/11/15 11:00, "cross-project-issues-dev-bounces@xxxxxxxxxxx on
behalf of Ed Willink" <cross-project-issues-dev-bounces@xxxxxxxxxxx on
behalf of ed@xxxxxxxxxxxxx> wrote:

Hi

Presumably you put tests in a separate plugin, so splitting off the
tests as a separate contribution gets you twice the limit with minimal
effort.

Perhaps a 10000 line limit might be appropriate for non-deliverable code
such as tests and build tools.

     Regards

         Ed Willink



On 19/11/2015 09:49, Sievers, Jan wrote:
Hi,

in the course of

https://bugs.eclipse.org/bugs/show_bug.cgi?id=477328


we had a contribution that slightly exceeded 1000 lines and thus
needed a CQ.
It took about one month to review it.

I am sure the legal team does its very best to keep up with the load,
so the following is in no way a criticism of the
people who actually do the legal review.

Rather take it as food for thought to whoever set up this rule.

IMHO the 1000 line rule is effectively setting the wrong incentives
for a thriving opensource project.

Here is why I think so:


The most diligent contributors add a lot of tests to their patch to
prove it works.
This is a good thing and we actively encourage contributors to
thoroughly test.
Test code can easily outweigh productive code being tested in terms of
LOC.
However this means the most diligent contributors, i.e. the ones you
want to attract, are more likely to hit the 1000 line limit.
Instead of thanking them for their hard work, we effectively punish
them with an extra month or more wait time before their patch can be
merged.
Apart from that, the 1000 line limit seems arbitrary to me because
technically you can split up any commit into any number
of smaller commits below the 1000 line limit.

Best Regards,
Jan





_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or
unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev

_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe
>from this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe
>from this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev
-------------------------------------------------------------
compeople AG
Untermainanlage 8
60329 Frankfurt/Main
fon: +49 (0) 69 / 27 22 18 0
fax: +49 (0) 69 / 27 22 18 22
web: www.compeople.de

Vorstand: Jürgen Wiesmaier
Aufsichtsratsvorsitzender: Christian Glanz

Sitz der Gesellschaft: Frankfurt/Main
Handelsregister Frankfurt HRB 56759
USt-IdNr. DE207665352
-------------------------------------------------------------
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev

_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev

_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev



--
"Have you tried turning it off and on again" - The IT Crowd
And if that fails, then http://goo.gl/tnBgH5

_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev


Back to the top