Now by moving to Github Pull request model in Apache Falcon, it creates a lot of pain as to each pull request merged would create a conflict in changes.txt for rest of the pull requests. Now considering this issue, the option of having the CHANGES.txt might have to be reconsidered.
On Sat, Jan 23, 2016 at 10:56 AM, Ajay Yadav <ajayn...@gmail.com> wrote: > I agree with you Venkat, mentioning in commit message is better than > mentioning in CHANGES.txt. Actually, in Apache Falcon we do both, hence > this is even more redundant for this use case. Although in my opinion even > mentioning contributor in commit message is weak form of attribution > compared to making the contributor the author of the commit. Attribution > use case will be perfectly solved with the github pull request model. > > It will also solve my biggest complaint about maintaining change log in > CHANGES.txt like this. The responsibility of maintaining the CHANGES.txt > will be shifted on multiple contributors rather than lesser number of > committers and active committers will not feel the pain. > > I have already created a JIRA to track this shift and work is already in > progress. > > > > On Sat, Jan 23, 2016 at 4:13 AM, Venkat Ranganathan < > vranganat...@hortonworks.com> wrote: > > > The maintenance of attribution is a valid thought, but different projects > > have handled it differently. For example, in Sqoop, it is written in the > > git commit message and the changes.list only lists the fixed JIRAs. > There > > are good things in both styles, but I prefer the Sqoop style for > > contributor information to be maintained in the Git repo as a comment. > > That makes the changes.txt easily readable for changes. > > > > That said, it is a preference set forth by the project team as Venkatesh > > Seetharam said. > > > > Thanks > > > > Venkat > > > > > > > > On 1/21/16, 10:35 PM, "Seetharam Venkatesh" <venkat...@innerzeal.com> > > wrote: > > > > >One more is to highlight incompatible changes which is very critical. > > > > > >We as PMC had decided to maintain this and I still think its very useful > > to > > >preserve it here. I do not want to use git to lookup. > > > > > >On Thu, Jan 21, 2016 at 8:38 PM Srikanth Sundarrajan < > srik...@hotmail.com > > > > > >wrote: > > > > > >> There are three main reasons for maintaining the changes.txt in my > view > > >> > > >> 1. Attribution of contribution to original owners > > >> 2. Live document listing changes by category (bug type) for > developers' > > >> convenience > > >> 3. A change log to refer to for someone who downloads binary/source > > >> releases contained the tar ball. > > >> > > >> @Ajay, It might be very helpful to call these out specifically. > > >> > > >> Regards > > >> Srikanth Sundarrajan > > >> > > >> > From: ajayn...@gmail.com > > >> > Date: Thu, 21 Jan 2016 13:24:21 +0530 > > >> > Subject: Re: [DISCUSS] Removing CHANGES.txt > > >> > To: dev@falcon.apache.org > > >> > > > >> > Thanks for your input Pavan, it is helpful to hear how everyone is > > using > > >> > CHANGES.txt and that is the whole purpose of this discussion. For > > >> checking > > >> > what all changes went after a JIRA, using git log is a better > option. > > For > > >> > example using CHANGES.txt you can not determine what all *features* > or > > >> > *improvements* went after this *bug fix *because there is no > relative > > >> > ordering between different types of JIRA in CHANGES.txt*.* Secondly > > you > > >> are > > >> > trusting that CHANGES.txt entries were always made in correct place > on > > >> the > > >> > top. Being an active committer I can tell you this is not true and > > often > > >> > during resolving conflicts etc. things get shuffled around. Also it > > is a > > >> > common error to put a 0.9 change in trunk while committing to > master. > > >> > > > >> > The whole idea is to automate the process, make it less error prone > > and > > >> at > > >> > the same time make change log available in better and more accurate > > >> format. > > >> > If hadoop script / workflow solves it then I am very happy to > discuss > > >> that > > >> > approach as well. If someone has a use case which can be solved only > > by > > >> > this workflow then I am happy to discuss how we can solve without > > >> > sacrificing those use cases but I request all of you to be cognizant > > of > > >> the > > >> > fact that as one of the most active contributors I am feeling the > > pain in > > >> > this approach and we need a better solution. I am open to listening > to > > >> > better ideas to solve these problems. > > >> > > > >> > If anyone has other use cases of CHANGES.txt file or any > > >> > opinions/suggestions then please chime in. Later, I will send a > > summary > > >> of > > >> > the discussion with all the use cases and a proposal to solve all > the > > >> pain > > >> > points which we can discuss further. > > >> > > > >> > *P.S.* We may be really used to the CHANGES.txt file but several > > apache > > >> > projects don't maintain CHANGES.txt e.g. SPARK, Flink > > >> > > > >> > Cheers > > >> > Ajay Yadava > > >> > > > >> > On Thu, Jan 21, 2016 at 10:50 AM, pavan kumar Kolamuri < > > >> > pavan.kolam...@gmail.com> wrote: > > >> > > > >> > > Not only for release notes if you just want to look what all > patches > > >> went > > >> > > after certain patch, it will be easy for anyone to just look into > > >> > > changes.txt and get it. Like suresh suggested we should think of > > >> writing > > >> > > script for generating changes.txt, that is good option. > > >> > > > > >> > > On Thu, Jan 21, 2016 at 8:42 AM, Ajay Yadav < > ajay.ya...@inmobi.com> > > >> wrote: > > >> > > > > >> > > > Venkatesh, > > >> > > > > > >> > > > I never said to point users to JIRA, I just said that > information > > is > > >> > > > available from JIRA and we don't need to maintain it in > > CHANGES.txt > > >> also. > > >> > > > We can extract from JIRA and put release notes and change log > > >> wherever > > >> > > we > > >> > > > want e.g. then we can choose to put it on site and putting on > > site is > > >> > > much > > >> > > > better than putting it in the code with no pointers to users on > > >> where to > > >> > > > see the change log. > > >> > > > > > >> > > > I am sure even you will agree that if we can get the change log > > >> without > > >> > > > having to update CHANGES.txt with every commit then it will be > > very > > >> > > helpful > > >> > > > for committers. We can just apply the same patch on the master > and > > >> the > > >> > > > branch with just 2 commands and no conflicts. > > >> > > > > > >> > > > > > >> > > > Cheers > > >> > > > Ajay Yadava > > >> > > > > > >> > > > On Thu, Jan 21, 2016 at 2:32 AM, Seetharam Venkatesh < > > >> > > > venkat...@innerzeal.com> wrote: > > >> > > > > > >> > > > > Ajay, I have made 3 releases on Apache Falcon and one in > Apache > > >> Atlas > > >> > > > and I > > >> > > > > beg to differ from your opinion. > > >> > > > > > > >> > > > > Pointing users to use Jira or Git for looking at release notes > > is > > >> bad > > >> > > and > > >> > > > > it immensely helps users to just glance at the text file to > > parse > > >> what > > >> > > > has > > >> > > > > gone into trunk. > > >> > > > > > > >> > > > > On Wed, Jan 20, 2016 at 11:38 AM Ajay Yadav < > > ajay.ya...@inmobi.com > > >> > > > >> > > > wrote: > > >> > > > > > > >> > > > > > Just to clarify I am not suggesting to not provide change > log > > to > > >> > > > users. I > > >> > > > > > am just suggesting to get rid of the practice of manually > > >> maintaining > > >> > > > > > CHANGES.txt file in the code. > > >> > > > > > > > >> > > > > > For example we can generate change log from JIRA(it provides > > >> release > > >> > > > > notes > > >> > > > > > for a particular release) and put it on website along with > > >> release > > >> > > > notes. > > >> > > > > > We can also consider the Hadoop method if it doesn't involve > > >> extra > > >> > > > manual > > >> > > > > > work for committers along with each commit. > > >> > > > > > > > >> > > > > > > > >> > > > > > Cheers > > >> > > > > > Ajay Yadava > > >> > > > > > > > >> > > > > > On Thu, Jan 21, 2016 at 12:58 AM, Ajay Yadav < > > >> ajay.ya...@inmobi.com> > > >> > > > > > wrote: > > >> > > > > > > > >> > > > > > > As I said Idea is to delete CHANGES.txt as the same > > >> information can > > >> > > > be > > >> > > > > > > extracted from JIRA. We can provide change log information > > >> using > > >> > > > JIRA. > > >> > > > > > > > > >> > > > > > > Maintaining CHANGES.txt file is a huge pain for the people > > who > > >> have > > >> > > > to > > >> > > > > > > commit patches. It is also very error prone way of > > maintaining > > >> > > change > > >> > > > > > log. > > >> > > > > > > We have a 6 weekly release cycle so that means that we are > > >> almost > > >> > > > > always > > >> > > > > > > running in two branches and it becomes very painful to > > >> maintain the > > >> > > > > > change > > >> > > > > > > log in this fashion. > > >> > > > > > > > > >> > > > > > > > > >> > > > > > > > > >> > > > > > > Cheers > > >> > > > > > > Ajay Yadava > > >> > > > > > > > > >> > > > > > > On Thu, Jan 21, 2016 at 12:38 AM, Suresh Srinivas < > > >> > > > > > sur...@hortonworks.com> > > >> > > > > > > wrote: > > >> > > > > > > > > >> > > > > > >> Hadoop has a script to generate CHANGES.txt. That might > be > > an > > >> > > > option. > > >> > > > > > >> > > >> > > > > > >> Agree with Venkatesh. This is an important information > and > > >> should > > >> > > > not > > >> > > > > be > > >> > > > > > >> deleted. > > >> > > > > > >> > > >> > > > > > >> On 1/20/16, 10:55 AM, "Seetharam Venkatesh" < > > >> > > > venkat...@innerzeal.com> > > >> > > > > > >> wrote: > > >> > > > > > >> > > >> > > > > > >> >I think this is quite useful and AFAICT, the release > > >> timelines > > >> > > are > > >> > > > > > >> >quarterly, it might be worth the extra effort to > maintain > > >> this. > > >> > > > > > >> > > > >> > > > > > >> >-1 from me. > > >> > > > > > >> > > > >> > > > > > >> >On Wed, Jan 20, 2016 at 10:52 AM Ajay Yadava < > > >> > > > ajayyad...@apache.org> > > >> > > > > > >> >wrote: > > >> > > > > > >> > > > >> > > > > > >> >> Currently we are maintaining CHANGES.txt to record > > >> > > contributions, > > >> > > > > > >> >> committers of the commits and the release in which > they > > >> got > > >> > > > > > committed. > > >> > > > > > >> >>All > > >> > > > > > >> >> this information is also available in JIRA and git. > > >> > > > > > >> >> > > >> > > > > > >> >> However, there are certain disadvantages of > CHANGES.txt, > > >> every > > >> > > > time > > >> > > > > > >> >>during > > >> > > > > > >> >> release we have to maintain different CHANGES.txt in > > >> master and > > >> > > > > > branch. > > >> > > > > > >> >>We > > >> > > > > > >> >> have to be careful with subtle details like "Proposed > > >> release > > >> > > > > > version" > > >> > > > > > >> >>and > > >> > > > > > >> >> "Released version" etc. This also creates confusion > when > > >> same > > >> > > > > commit > > >> > > > > > >> >>gets > > >> > > > > > >> >> committed into master and the branch. > > >> > > > > > >> >> > > >> > > > > > >> >> Also, it entails committers to do some edits to the > > patch > > >> > > > submitted > > >> > > > > > by > > >> > > > > > >> >>the > > >> > > > > > >> >> contributor. This is error prone and can be tedious > > >> sometimes. > > >> > > > > > >> >>Sometimes we > > >> > > > > > >> >> forget to attribute contribution or spell > contributor's > > >> name > > >> > > > > > >> >>incorrectly. > > >> > > > > > >> >> > > >> > > > > > >> >> Hence I propose to delete CHANGES.txt from master > (0.10 > > >> > > onward). > > >> > > > > > Please > > >> > > > > > >> >> provide your inputs. If everyone agrees, then I will > > >> create a > > >> > > > JIRA > > >> > > > > > and > > >> > > > > > >> >> delete CHANGES.txt from master. > > >> > > > > > >> >> > > >> > > > > > >> >> > > >> > > > > > >> >> Cheers > > >> > > > > > >> >> Ajay Yadava > > >> > > > > > >> >> > > >> > > > > > >> > > >> > > > > > >> > > >> > > > > > > > > >> > > > > > > > >> > > > > > -- > > >> > > > > > > _____________________________________________________________ > > >> > > > > > The information contained in this communication is intended > > >> solely > > >> > > for > > >> > > > > the > > >> > > > > > use of the individual or entity to whom it is addressed and > > >> others > > >> > > > > > authorized to receive it. It may contain confidential or > > legally > > >> > > > > privileged > > >> > > > > > information. If you are not the intended recipient you are > > hereby > > >> > > > > notified > > >> > > > > > that any disclosure, copying, distribution or taking any > > action > > >> in > > >> > > > > reliance > > >> > > > > > on the contents of this information is strictly prohibited > and > > >> may be > > >> > > > > > unlawful. If you have received this communication in error, > > >> please > > >> > > > notify > > >> > > > > > us immediately by responding to this email and then delete > it > > >> from > > >> > > your > > >> > > > > > system. The firm is neither liable for the proper and > complete > > >> > > > > transmission > > >> > > > > > of the information contained in this communication nor for > any > > >> delay > > >> > > in > > >> > > > > its > > >> > > > > > receipt. > > >> > > > > > > > >> > > > > > > >> > > > > > >> > > > -- > > >> > > > _____________________________________________________________ > > >> > > > The information contained in this communication is intended > solely > > >> for > > >> > > the > > >> > > > use of the individual or entity to whom it is addressed and > others > > >> > > > authorized to receive it. It may contain confidential or legally > > >> > > privileged > > >> > > > information. If you are not the intended recipient you are > hereby > > >> > > notified > > >> > > > that any disclosure, copying, distribution or taking any action > in > > >> > > reliance > > >> > > > on the contents of this information is strictly prohibited and > > may be > > >> > > > unlawful. If you have received this communication in error, > please > > >> notify > > >> > > > us immediately by responding to this email and then delete it > from > > >> your > > >> > > > system. The firm is neither liable for the proper and complete > > >> > > transmission > > >> > > > of the information contained in this communication nor for any > > delay > > >> in > > >> > > its > > >> > > > receipt. > > >> > > > > > >> > > > > >> > > > > >> > > > > >> > > -- > > >> > > Regards > > >> > > Pavan Kumar Kolamuri > > >> > > > > >> > > >
