Articles on taapasX28's Blog

Final Checkin

taapas2897@gmail.com (taapasX28) — Sun, 25 Aug 2019 14:10:41 +0000

Hey everyone! This is my final blog post here. GSoC 19 has finally come to its end. This week I mostly worked on hg abort --update

What did I do this week?

This week I worked on writing intensive-tests for hg update --abort
As suggested by @mharbison72 I worked on cases where we need to take considerations of subrepos and check cases where a subrepo is dirty and cases where more than one subrepo is present and conflicts for one or more of them is resolved but not all conflicts are resolved so as to check the unmark feature. The patch is still under review but it too will soon get accepted.

Did you get stuck anywhere?
I did not get on anything major this week . I did have to perfect the code considering some endpoint cases but it came around fine enough.

Finally, I would like to thank Pulkit for reviewing the patches and solving the doubts despite having a busy schedule, Martin and other people in the community for teaching the basic etiquettes and facts required to be a better open-source contributor and solving the doubts no matter how silly they were.

Adios !

Coding Period: Week 12

taapas2897@gmail.com (taapasX28) — Sun, 18 Aug 2019 14:09:13 +0000

Hello everyone! This is one of the last blogposts. This week I worked on the last features I needed to add in my GSoC period i.e. hg update --abort

What did I do this week?
As stated in the previous week I worked on hg transplant --abort. As suggested by @Pulkit it was modified to --stop flag and finally got it merged [1]. I also resolved most of the hg continue series of patches.
Later this week I was finally able to deduce a logic for hg update --abort and sent a patch for that [2]. It is still under review though.

What is coming up next?

As almost all patches are dealt with I will modify the patch for hg update --abort as suggested by the community and get that merged.
I will also the continue patches for evolve at the beginning of the next week and get them merged too.
This issue has been one of the most requested features and I am really happy that I could work on this and get it resolved. Also in this week, I will be working on my GSoC 19 blog which is supposed to be attached as prescribed by final product submission guidelines.

Did you get stuck anywhere?
I got stuck regarding the logic of hg update --abort for long enough but a deeper understanding of mergestate and associated functions help me deduce a logic which was not functional initially but later this week I was able to make it work. I am still adding more extensive test cases for it to make the feature perfect. It is one of the most interesting features that I have worked on this summer which involved developing a deeper understanding of the merge and update workflow.

Coding Period: Week 11

taapas2897@gmail.com (taapasX28) — Sun, 11 Aug 2019 17:41:45 +0000

Hey everyone!! This is what I did this week. In this phase, I worked mostly on getting my previous patches on hg continue merged and on hg update --abort.

What did I do this week?
This week I was asked to make some modifications on the patches that I previously sent regarding hg continue . I had to refactor things for graft and clean up the code. Further, I had to modify the code for transplant too. I also worked on hg transplant --abort and it is still under review.
The code for hg update --abort is still under progress.
What is coming up next?

This week I am planning to get hg continue series and hg transplant --abort merged. Also, I would like to get hg update --abort to a decent review level.

Did you get stuck anywhere?
This week I did not get into any major problem. While refactoring code for graft I did face the problem of not being able to remove opts as the argument from updateopts function. However, I later realized that opts were not being loaded properly which was causing the bug . I will try to resolve this bug this week.

Coding Period: Week 10

taapas2897@gmail.com (taapasX28) — Sun, 04 Aug 2019 16:22:45 +0000

Hey everyone! This is what I did this week. In this phase, I worked mostly on the understanding of mergestate for developing a patch for hg update --abort. This is the last part that I need to complete before GSoC deadline.

What did I do this week?

As stated previously I worked on evolve and was able to get my patches for hg abort for evolve repository merged which included adding support for hg evolve and hg pick. Also, I worked on hg continue support and was able to add support for evolve. I will submit the patch for that once pick is also complete.
As for hg update --abort I went through the code for hg resolve -u and
also -at:local. I learned what merge state stores and started writing the code adding --abort flag for hg update.
I also received reviews for the continue patches I sent to the core repo and modified my patches according to that.

What is coming up next?

This week I am planning to complete support for hg continue for evolve. Also, I would send my first patch for hg update --abort.

Did you get stuck anywhere?

I got stuck with the logic of abort flag but @Pulkit pointed me to go through the code and documentation of mergestate which helped me get a better understanding of the workflow.

Coding Period: Week 9

taapas2897@gmail.com (taapasX28) — Sun, 28 Jul 2019 19:52:18 +0000

Greetings everyone! Update for this week is as follows. In this phase, I sent the remaining patches required to complete hg continue support for core Mercurial repository.

What did I do this week?
As stated in the previous week I sent patches for hg continue for for graft, transplant and histedit. [1] [2] [3] [4]
I also created logic for hg transplant --abort[5] as asked by @Pulkit and added its support to hg abort plan [6]. These patches are still to be merged and under review.

What is coming up next?

This week I am planning to complete support for hg continue for evolve. Also, I will start my work on logic for hg update --abort which is one of the most demanded features for Mercurial.
My patches for hg abort for evolve merged as they still left to get reviews as most of the people in the community are busy with the new version release.

Did you get stuck anywhere?

I did get stuck when I started with the logic for aborting a transplant but @Pulkit again helped me out of the situation and suggested that I should look into the recover method of the transplanter class.

Coding Period: Week 8

taapas2897@gmail.com (taapasX28) — Sun, 21 Jul 2019 14:27:37 +0000

Hey everyone! This is this update for this week. This blog describes basically how I started working on hg continue and got my first four patches approved.

What did I do this week?

This week I started my work on hg continue and sent the first patch regarding the logic for it [1]. Then subsequently I sent patches to add support for rebase[2] and unshelve[3] and also for graft and it was queued but as it wasn't too refined it had to be de-queued and sent the fix[4]. This leaves graft, tranplant and `histedit` to work on to complete hg continue plan.

What is coming up next?

This week I am planning to complete support for hg continue for graft, transplant and histedit and get that approved. Also, I will try to apply it move to evolve.
I also am still planning on getting my patches for hg abort for evolve merged as they still left to get reviews as community is busy with the release.

Did you get stuck anywhere?

I did get stuck when I started adding support for the functions but @Pulkit again helped me out of the situation. I was not sure how to pass on the opts for various commands when continuefunc is directly called . But the knowledge that opts are present in state file resolved it.

Coding Period: Week 7

taapas2897@gmail.com (taapasX28) — Sun, 14 Jul 2019 19:35:39 +0000

Greetings everyone! This is this update for this week. This is all about how I
got hg abort finally added to Mercurial. I also sent PRs to evolve repository adding its support.

What did I do this week?

This week I improved hg abort further and finally got them accepted and merged. This marks the end of hg abort plan for Mercurial core repository.
It will now be shipped with Mercurial version 5.1. It is still marked EXPERIMENTAL though as there is still a little doubt regarding its perfectness.
Also, my patch for bug code 6160 was finally merged. Completing integration with the core I moved to Mercurial's evolve repository and added support for hg abort for hg evolve --abort and hg pick --abort [1][2][3]. These patches are still under review.
I also started my work on hg continue and was able to develop a logic for it and also apply it for histedit.

What is coming up next?

This week I am planning to start sending patches on hg continue on Phabricator. Also I will try to apply it for the remaining commands in core and then move to evolve.
I also plan on getting my patches for hg abort for evolve merged.

Did you get stuck anywhere?

I did get stuck a little when working with evolve repository but @av6 and @marmoute covered most of them for me. Besides that I wasn't stuck in something that was major.

Coding Period : Week 6

taapas2897@gmail.com (taapasX28) — Sun, 07 Jul 2019 17:07:15 +0000

Hello people! This is this weeks update. This week was spent mostly in the review of hg abort. It has now come up to be much refined and acceptable than it initially was.

What did I do this week?

This week I spent most of the time improving hg abort patches according to the suggestions of @Pulkit [1] [2] [3] [4] [5] [6] [7]. As the new release nears I was asked to mark it EXPERIMENTAL for now so that the related bugs can be dealt with before it is finally shipped with Mercurial.
I also worked on bug code 6160 [8]. Since it was very much related to the field I was working on so I was asked to fix it and I created a patch for that [9].

What is coming up next?

This week I am planning to start my work on hg continue and start sending the patches for that. I will also try to get hg abort and my patch on issue 6160 committed this week.

Did you get stuck anywhere?

I did not get stuck with anything major this week. Small minor doubts did arise but the community and my mentor @Pulkit helped a lot.

Coding Period : Week 5

taapas2897@gmail.com (taapasX28) — Sun, 30 Jun 2019 20:25:39 +0000

Hey everyone ! This is the update of what I did this week. This week I got my API and its extended cases committed. Also, I completed the implementation of hg abort and pushed the patches for review.

What did I do this week?

This week as stated previous blog I completed the implementation for hg abort and included shelve and merge in the plan. Also, I refined my patches and send them to Phabricator in a well-divided format.
In this period I also got my state detection API committed to the Mercurial repo. It's extended cases suggest by @martinvonz were dealt with and it was my patches were finally merged. I also updated the release notes regarding that. This marks an official end to the first sub-task that was the task of the unification of state detection APIs.

What is coming up next?

The coming week I am planning to get all of the patches for hg abort approved and update the patches according to the reviews. There might be a lot of new features that might be needed to be added to that as hg abort is still in its underdeveloped stage.
I will also work parallelly on hg continue and see if I can prototype it.

Did you get stuck anywhere?

I faced some problem when I was trying to include merge in the hg abort plan. Though it took some time, I was able to figure out the workflow and make an appropriate patch for it.

Coding Period : Week 4

taapas2897@gmail.com (taapasX28) — Sun, 23 Jun 2019 12:45:44 +0000

Hey everyone!! This is the update for week 4 and hence the last week before the first evaluation. This week I mainly extended the state detection API to more cases which were not part of the plan but an essential for Mercurial. I also came up with the first implementation of hg abort.

What did I do this week?

As stated in the previous blog I got my PRs regarding the first sub-task accepted.[1][2][3][4] @martinvonz suggested some extended cases for the API and I was interested to take that up and add those features. Also, I updated the release notes about the new API.
For hg abort, I sent the first prototype for it which included the algorithm for hg abort, the extension of state detection API to register abort functions and inclusion of graft, rebase and histedit into the plan. This leaves me with two remaining operations merge and shelve. I was able to write a prototype for shelve but since it was buggy I did not send it as a PR. I got my patch reviewed by @Pulkit on Bitbucket [5]. He suggested me some changes I will send the patch directly to Phabricator.

What is coming up next?

The coming week I am planning to get all of the patches for extended API approved and will send the complete hg abort patch to be reviewed. The work that is still left is cleaning up the patch and inclusion of shelve and abort as part of abort plan.
Towards the end of this week if time permits I would like to start with hg continue.

Did you get stuck anywhere?

During the start of hg abort I did get stuck with what actually will the algorithm for it look like. Though I had a visualisation but was not quite clear with it. I discussed this with @Pulkit got my doubts cleared and went ahead with my work.

Coding Period : Week 3

taapas2897@gmail.com (taapasX28) — Sat, 15 Jun 2019 20:27:54 +0000

Hey everyone ! This is the update of what I did this week. This week I tried to perfect my implementation of things and learned how to write code in a minimalistic way instead of long complex hunks.

What did I do this week?

This week was mostly spent on perfecting the unification code whose PRs I sent last week. I improved them and added a couple of new features. The prominent one being writing a method to register to unfinishedstates. I also worked on the unification of merge to checkunfinishedstate() method which would be a large change to the whole codebase as previously separate method per command were used to track down a merge in progress while the command was under execution.
I parallelly also tried to register abort methods for various commands in the core into the unfinishedstates list and develop a miniature hg abort command. Its support needs to be extended to extensions and then it would be ready for review.

What is coming up next?

The coming week I am planning to get all of the patches for API unification approved and also will try to send some patches regarding hg abort.
I will also work parallelly on the unification of mergestate to checkunfinished() because the unification of that would require addition of certain functionalities one of them being providing custom hints as per the command.

Did you get stuck anywhere?

While unifying the merge with other unfinishedstates I found the tests for hg strip were failing and the changes in them seemed quite illogical. I talked to my mentor about it and to much of a surprise he too found out that it was a bug in strip that it was allowed even when a merge was in progress. I prepared a patch for that and fixed that.[6]
I also got stuck at the initial reviews of the patch which demanded merge and bisect to be registered at the end of the list. However, my mentor's suggestion on developing a registering mechanism came handy and that issue too was solved.

Coding Period : Week 2

taapas2897@gmail.com (taapasX28) — Sun, 09 Jun 2019 13:38:17 +0000

Hello everyone !! This is the update of what I did this week. This week was full of new learning experiences and a lot more productive than the last one.

What did I do this week?

This week completed the implementation of unification of APIs subtask and uploaded my patches on Phabricator[1] [2] [3] [4] [5] for final review. As I stated in the last blog, most of my doubts were resolved last week which gave me a headstart to proceed further. I was able to complete the subtask but I also needed to clean up the code and make it presentable on Phabricator. Initially, I uploaded the whole code at once which was quite untidy and difficult to review. Martin suggested me to break it down into simpler patches which was a little frustrating at first but later gave me a really important thing to learn that one should not only code from the perspective on an author but also from the perspective of a reviewer. Because in later times too if someone needs to check out changes that the code made, he or she may do so with ease.

What is coming up next?

The coming week I will move according to the timeline in my proposal and start my work on hg abort and make it compatible with core commands like hg graft.
Also, I will be modifying my patches on the unification of APIs and get the patches merged. One of them has already been accepted.

Did you get stuck anywhere?Did you get stuck anywhere?

During the starting of the week I was doubtful regarding some issues like
does checkunfinishedstates() be the part of the new class I created or not and so on. But my mentor Pulkit cleared most of them.
Later while splitting up my patch into smaller patches I was stuck in a condition where I needed to make an amend to a commit that was not the very recent one. I could not figure out the workflow and asked my mentor about it. He gave me manual of hg evolve to go through saying that would make my workflow more powerful. After going through it I realised how awesome Mercurial is and hg evolve is one of the coolest feature that I have seen in any VCS. It provides with a whole new level of flexibility and also I was able to learn how things work at changeset level when hg amend or for that matter and history editing function is used

Coding Period : Week 1

taapas2897@gmail.com (taapasX28) — Sun, 02 Jun 2019 17:58:25 +0000

Hello everyone !! This is the weekly update of what I did this week. This week was full of ups and downs but in the end, was quite fruitful.

What did I do this week?

This week as stated in my previous blog I tried to make a rough implementation of the code. I unified the state determining APIs and merged their functionalities w.r.t state determination. I pushed the code to my clone at bitbucket[1] [2]. However as stated by my mentor, though the unification was alright it required to be a more generic implementation and a few more functionalities must be added to it which would help me in the implementation of hg abort and hg continue. I had a video call with my mentor Pulkit today and he helped me clear my doubts and gave me an approach on how I could improvise my code. He asked me to proceed in steps and check for different implementations and then decide what was suitable for the task. He also advised me not to stay stuck with something for very long and ask questions if stuck.

What is coming up next?

The coming week I am planning to make my API more generic and finish it off as soon as possible so my mentor can review it and I can make the necessary modifications. Towards the end of the week, I also plan to start with the implementation of hg abort given this task is complete.

Did you get stuck anywhere?

During the starting of the coding period, I was a little unconfident about my implementation part but eventually, it did work out. Also, I got stuck with the values of allowcommit flag for different commands in hg but my mentor Pulkit was there to help me out of the problem and pointed out the part of the code I should read to get the appropriate values. Most of my doubts were cleared this week and though the implementation was partially correct it gave me hint how to proceed. This week I plan to correct my wrongs and proceed forward with greater speed.

Introductory Blog

taapas2897@gmail.com (taapasX28) — Tue, 28 May 2019 19:45:43 +0000

Hey everyone! I am Taapas Agrawal, a sophomore at Indian Institute of Technology, Kharagpur. I am working under Sub-Org Mercurial, a distributed version control system. My project is about implementing hg abort and hg abort commands which are currently absent in Mercurial. This is what I will be working on for the next 12 weeks as my GSoC project.
My project is basically divided into three major tasks each having a variety of subtasks. The three major tasks are as follows:

Unification of the state determining APIs.
Implementation of hg abort in core as well as extensions.
Implementation of hg continue in core as well as extensions.

Completing these 3 tasks timely would lead to the completion of my project.
What did I do in the community bonding period?
During the first week of the Community bonding period, I had a 45 minutes video conference with my mentors which was basically to give a kick-start to the project. Many of my initial doubts were dealt with in the meeting and I was given a very good starting point from which I could start determining the solution of the first major task of Unification of APIs which in my view is toughest among the three. I was asked to send a mail in the Mercurial mailing list in case I have a major doubt or ping on IRC.
During the period I analysed the hint given and was able to understand the code better. I read through the code of the two APIs and tried to make a list what all the features and functionalities individually each of them has and how a combined API could be made to perform those tasks.
I asked any kind of doubts on IRC and the mentors were really helpful and gave me suitable hints. Towards the end, I was able to write a rough pseudo-code of what the new API will look like and started writing code for that.
During this period I also planned out how I will implement the other 2 major tasks of implementation of hg abort and hg continue.
What is coming up next?
The coming weeks I am planning to complete the code for the new API and get it reviewed by my mentor before actually merging it with the rest of the code. The patch will be buggy so I plan to remove these bugs as soon as possible with the help of my mentor.
Did you get stuck anywhere?
There were instances where I could not figure out what some pieces of code do and I came across functionalities which the two APIs had and that could not be combined as it is but require a totally new method and a new way of implementation.