Jenkins Google Summer of Code Office Hours, 20 Mar 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Jenkins GSoC Q&A for students. Mar 20, 2019

Description

https://jenkins.io/projects/gsoc/

A

We do, sir again.

B

A

Yeah, my article, yes, you are okay, so previously I asked the same question with Martine. Sir I just wanted to I just wanted to show you my UI vision, because I want you to discuss about the UI, how it's going to implement.

B

So excuse me: let's cut from the process of specific questions and then yeah, then we will switch to project specific ones. Okay,.

A

You're asking me right like it was: it is related to the project itself.

B

So, let's just spend some time on whatever organization's.

B

A

A

Okay, so I wanted to know exactly you. A pattern of the project, artifact promotion plug-in so I'll share my screen a.

A

Little bit Devil's knot. Yes, yes, okay, so I just want you to ask that this new item section that was here when the promoted that the new plugin is implemented, then, where would it be placed like like? Will it be placed alongside these projects freestyle projects? There will be something like promotion promotion of artifacts. Would it be placed at this particular place or like a the job? We were actually creating the job.

A

Yeah I knew right, so would it be placed alongside these projects or.

B

Us so with support of the design, you can do it in different ways. My personal understanding first pitch would be just the subtype of pipeline, so we a it, would appear in this list unless you.

C

Wouldn't just be a property of a job pipeline, not a special type.

C

Somebody if somebody is tapping their microphone.

B

So there are two parts of the question: one is how it's being defined another one. Oh it's being so yeah. It would be great if we could define promotions and promotion logic directly pipeline definitions or in declarative pipeline special sections, but you mind listening. It would still appear as a separate job type management page, since this is.

A

Being mentioned that there should be a squad on standalone plugin right, so it should see in this piece and next question would visa like, whenever I go inside a certain project and go inside a particular bill right so inside the computer section, we have this section right for bills when this section this will be migrated to that job right. The new job.

B

As execution or as definition, because so you can still define this promotion process inside your pipeline, if you want.

A

I can want to define this in the pipette or I can make a comment on that. What you I.

B

Mean sure I get your question.

A

This thing that you want bill to in this section would be migrated to do a new job if I'm not wrong. So.

B

You can define it in both ways, depending on the preference you can define it inside existing pipeline is a property like flow suggested or is a special clothes, for example in the credit pipeline or you deployment in external job. So it's a matter of definition, and you know they expect you the student to come up with proposal. They're. Both options are technically possible and both options. Many year find some users.

A

Okay, I do like both of the options are fine I can implement any of them. Yes,.

C

You can, can you elaborate on the the reason to create a new type there so.

B

There is no reason to create a new type of job. The reservation, our general reason to have promotions as separate enjoyment. Ideas so like we were discussing yesterday. I believe promotions currently are not well mapped into Jenkins job, our conception.

B

So what we would like is we would like to have promotions as separate job entities which become instruction so, for example, there may be a folder or multi branch pipeline and inside this entity there is a common job definition, so you don't use hogs as item groups which I represented a job as job properties, but this is an architectural thing, even if we talk about job types etc.

B

No I do not think that it has to be a job, a new job type because yeah, if you talk about promotion support for pipeline, it would be rather expected that the promotion is also pipeline. It's.

C

A property like you cannot define in there either there as an automatic promotion or a manual promotion. Yeah.

B

Right but inside it will be still a pipeline code, it would be still executed as a separate job entity or maybe as a separate run entity depending upon how we implement that, but it would be just a part of third party such thing definition as I tried to say it may be different. So the finish of promotional logic. It may be set your original job. It may be defined in external job. It depends on how I.

C

Could see the view the view part as a separate entity is kind of because it goes across the whole, the whole Jenkins, whereas the execution would be an extension of the queue and architecture. Does that make sense? Maybe.

B

Well, I think that you should really plan a secret meet you for this visit, no project so that we could give director projection. So maybe, if you guys have some time, we should take it offline and schedule a meeting like all the projects.

A

So my next question is like, as your guidance I went, I read the release blog plugin, since it's compatible with the pipeline, so I just read through all the codebase and what I found is one in one of those classes inside the release, plugin like the released promotional conditions or Java in this class. What it does is, it extends a promotion condition which is which happens to be the class of promoted pipe promoted, build plugin.

A

So this means that all the most of the classes which are inside the promotion bill are compatible with the pipeline, but there are some which are not so. How do I distinguish between those which are compatible to those which are not line? Go for.

B

Form so that if you see classes like epic project or abstract build in the course, you can expect that this class or these methods you know compatible this pipeline.

A

Okay, like promotion, dot, java promotion process, dot java job property. These are not compatible because they've got abstract classes like abstract builds. That's why you know migrate. Ok, so that's right! Ok! Ok! Next question is like if abstract builds are not supported by pipelines. If abstract projects are not supported by pipelines, then what are the? What is the actual project type which is compatible with pipeline, which is supported by pipeline.

B

Tricky thing because pipeline is a separate project type, so it may be useful for the students as well probably I'll just share my screen. If you don't mind, okay,.

B

Yes, so probably yeah, let's just try to turn box of it.

B

Okay, so job is kind of top-level entity for executable types, so job has methods like so yeah it generally contains runs, and this runs somehow they been. So there are many implementations of jobs. If you see here, you may see that there is abstract project which is a part of job. So it's one of the communications of jobs. Another implementation is pipeline, so on the same level, there are other project types, for example, inhabitants, project or other projects which inherit the job type.

B

So what it means that this project types are really independent, and these are part of the tree, although a base shade this job in their face and also all courses up to the hierarchy, so you may see that job implements abstractly and also that they, it may be a part of the item tree inside the changes that is actionable, so it may contain ocean and make it c-sections and yeah.

B

This is but job is the basic definition which just just applies for every job type and Jenkins same around here for answer that is, the base applies all drum, which is exactly the same.

A

Beliefs over which exact bill type would be supported from in the pipeline of the job type.

B

Pipeline because pipeline is a Septon job type. Okay, if you see a use keys for supporting jobs in general, so, for example, supporting freestyle job as a promotion logic for verifying. Then you can definitely do that in your proposed. If you feel it's reasonable, but yeah expectation would be that at least pipeline promotion logically supported.

A

A

Next question is there's in the in the project proposal its mentioned. It has mentioned that detach detached the existing promotion, condition. Extension point and implementations from promoted builds plug-in. What does that mean so like getting rid of all the existing extension points of the plugin? Well, so.

B

What it means is it been common knowledge except she think. If it, there are many EPIA plugins, so plugins, which provides shade functionality and being reused by other plugins, for example. Promoted bills already includes extension points for defining promotion conditions, example, manual, promotion, condition when somebody quits a button or dependents or whatever other event and Jenkins. There are maybe a dozen of implementations for this promotion condition, so they do this line in.

B

But if you create a API plugin and if you move for very usable functionality out of the promoted deals when other plugins and consumed its functionality without including the rest of the promotion fields code. So the idea of this body that very you just detach the plugins then effectively reuse the usable components of existing.

A

So for the for the for the code, part which is to be implemented by the pipeline, I should mention the extension point in it right like in a file booth, the class which is along with the at the rate extension point. It is read by the pipeline. That's the exact mechanism of how pipeline works right, sir.

B

A

So last questions are, like you mentioned in a vision in a vision that, in the properties section we must include, we can improve the environment like where, where build is going to be deployed, and can we do that by using the existing fingerprint that is being used in the promotion? Build already me.

B

Yes, you can okay.

C

What do you mean by deployed in this in this case here, meaning.

A

Like certain bits, after being promoted, they would be deployed into some sort of the next stage. They would be passed to the next stage so just to track where it is being deployed. Can we use fingerprinting? That was my question.

B

Yes, all fingerprinting can be used for tracing of any dependencies between fields and between any kinds of artifacts and within Jenkins, and you can use it here to track dependencies. Okay,.

A

So last question last question is like what you like: if I need to refactor the codes, the classes which are not compatible to buy things, what would what should be made? What should be my general approach? What should be my general approach to refactor them so that I can make them compatible with the pipeline, like abstract, like I, need to replace that with some other build type or like promotion? Job I need to change that. What should be my general approach to refactoring those codes, yeah.

B

So again, general practice, this project that we interested to retain compatibility where possible. So if you see that we need to perform it into the factory, there are ways to written binary compatibility. There are special week page, which explains how to do that and if you work on any project, and if you need to do a massive change, it might be a place.

B

You want to start from so Jenkins binary compatibility, I believe, let's see an actual inside retaining working silicon, so this would be one of the starting pitchers from for you, because it references several rejected, specific ways, making a trivet number one. You change the detail when you remove a few to remove a method but yep. If you see any significant change in P, is definitely something which will require significant design and before doing that, yes see whether there are alternate options, it doesn't apply to only this project. It applies to all projects.

A

Like in one of the tutorials in the plug-in development, I came across the fact that variable substitution, that is not enabled by plugin, so there was an idiom that was given so by including that it can be made compatible with both form of projects from from pipeline and freestyle. It mentioned something like that about the idiom of extension points. I, so would that be the exact tutorial.

B

So yeah, when you have a specific question, you may see that it's not really effective to ask for them on the call, because sometimes it's hard to express what you need and that's why we recommend using maybe increase or give her shots for that, because it may be more effective in many cases. Okay, thank you. Thank you. So what I propose to just take it offline because yeah we have a limited time.

A

B

Have other students all.

A

B

So if you have some time after the call- or we can return back to that but yeah, my proposal is really writing questions and probably next time we will just try to turn both questions for five minutes and we will be rotating between students, cuter.

A

B

Okay: okay, when.

A

You think again.

B

Let's see if we can discuss something between them, so.

D

Here's a connect in a shamash way right now.

D

D

Wait a minute wait. So there are some quite she's concerned. Was it aware CFS support so basically in the cloud storage for expenses, I found some really like confusing points of this idea. That is, the urgent problems was aw CFS report. The first one is the only ones we is easy to instance.

D

So actually it is highly integrated with the existing in our us easy to, and the second one is I think either this once their user mom Mary EFS too easy to actually when I, when I create an ec2 instance and I only need to run one command line to one of those what ifs to12 Jenkins do the job for the user. It might be much harder in the users do itself, so there might not be much benefits to the user if they have to use engine.

B

Talk about use kiss when you provision an engine, an engine fails efis out of the box. Then yes, you probably don't mean to implement something specific, but imagining doing it. Provisionary efis on the month within pipeline, for example, via pipeline richards, whatever my step, testing state stage of coverage and then needs on data in such case, you can use if se here is with more data on a machine, but it in traditional these pieces. Another use case is still the same. We face instances between machines can be also done a inside workspace panel for you.

B

So, for example, you need to first work space between multiple agents and you use. It is implemented.

D

A

D

You can continue so.

B

I just provide my examples and again, if s, there is just an example. If you don't see much benefit Oh for using EFS as a reference implementation, you might think about other implementations, because even if it was just by worried because one of the examples yes.

D

Yes, actually I have to discard through something interesting during the exploration of the universe, yet so I think that it is doable for AWFS and there are an existing ýà FS support that a is that easy to plugging so basically I think the most hard the hardest part of implementing the EFS support is I need to get access to the users. Access rights right, I need to make my Jenkins as a user, so I need the credit user user names or something like that to log into their console.

D

I think this is most difficult part, but it's already done in the ec2 plugins, it's just in the easy to blogging. This is one we used plugin. So in this is a plugin. It has already implemented a ways to get access to the console and then sure that console we can send the API. We can send in a request AWS, and so that includes create an ec2 instance, long changes and maybe do other things that a normal user can do. So.

D

Basically, if we will want to implement the EFS, we can make you some of the code of the easy to plug in I. Don't know if that is doable.

B

Justly you may want to deploy dependency on easy to plug in directly so yeah, as we discussed before. There are extension points and Jenkins. This extension points may be mandatory. They may be optional, so, for example, you can declare optional dependency from UC to plug-in, with optional extension upon documentation so that you can still use external workspace manager plug-in without easy to, if needed.

B

All there may be other ways, for example, moving shade functionality from PC to to another plug-in like epi plug-in example, there is a plug-in called AWS densha's which allows managing for a surprise, AWS credentials. So maybe something could be done like a bit for EFS.

B

Just one of the examples. Oh.

D

B

Yeah, if you really want to focus on EFS in your project, then yeah- it's nice, so that you just have this easy to plug in the recommendation is to think how you integrate the things we each other. It would be great for your project proposal if you explore this area and see how they do together.

D

Okay, so actually this another idea here, if that I think that I'm every. If you want to support two different cloud service providers, we need to offer an API plugin for those of all the for each of the cloud providers. I think that is a better practice. Better line hey if any other plugins wants to use if I might just need to access the cloud providers.

D

So as you mission- and there is a 8mc credential plugin, so I think that might be really helpful for this job, because I can just use it and then I can use something like maybe a dummy console to send requests to the in episode. @B wouldn't be more doable for me.

D

If so, basically, I think that in some way, so here's at least another project idea so and as a mission name, the original, prod project ideas I think there will be some cloud service providers even consider like, for example, a lot about a cloud is a very good, so service provider, training, so I think for as a Chinese I, don't like to implement something for the Alibaba and for me, I think the best way to support external workspace manager with a specific cloud service provider and better to first of all, like I, have to implement it on cloud service provider, ap I plug in, and then you sub, like your login to the cloud service provider and then issues requests.

D

I think this would be a more doable design for for all the kind of things you like it, the couple: what do we do in the external expect manager and login to the service provider? What do you think yeah.

B

It might be feasible design and if you want to do such things, you could also take a look at the J clouds. Now there is a project called Apache j, qf j cloud.

B

It's effectively a set of libraries for what you need is different cloud platforms using fresh unify KPIs mm-hmm, it's the burning from the chat, so it might be one of items you would like to take.

B

If you want to go with this, unify DPS idea and.

D

So here is a discussion. Is over I like to want solid ad, on which one we really want to do, because in the proposals I'm not sure which one is a better. So if anyone can give me some suggestions like how do I gonna implement, for example, AWS EFS support in the external space manager and from the choice is a mission above.

B

B

It's something maybe.

E

I'm, not very sure I did understand the question. So can you like quickly repeat please so.

D

Basically, there is a to waste. You support it obvious. If support the first one is I. Do I write the code that I'm logging to the cloud service provider that is I, become Oh, a user of the ADA base and write this part of code in external, your base and that's design? But if we want to support multiple cloud providers, the cloud base of the external expenditure will become large and the other solution is I. First write a like aw as I, see it abyss or some providers API plugin.

D

So that offers me the function to looking other service. I shall act as it called user and then I can use a little those API plug-in to send actual requests. Some specific requests to the cloud providers, the mething- that is a more clean design of it, which one do you think is better yeah.

E

I also agree with you, I think. The second one is clear because you actually decouple the logic from from the xws plug-in with the actual login and / and credentials and stuff like that in in another plugin, but I'm wondering if you can integrate this somehow with what oleg mentioned earlier with the AWS credentials, plug-in or stuff like this I'm, not very sure if they can be integrated together somehow so so so that you don't need to actually write another plugin.

E

You can just rely on that plug-in, do to use the credentials for the AWS, but but the main idea is that separating these I think it would be a better idea than just having the code inside the external workspace manager plug-in.

D

Yes, pretty obvious: yes, I might have already an existing plugins to use, but for other cloud providers we value Bob, Emilia zero or something like that. I might need to write for write different, maybe I need to write different plugins or the one that the Oracle, what I'm not sure what is ever just as automation. It looks like a bee camera a uniform and if you had to use all the cloud service providers, so I think that would be after investigation.

D

I can maybe a clear it in clarify in my proposals so and here's another question so can I just write it? Those are the same proposals in the class space, a cloud storage and support for external space. So in this project right, here's one do I need to write an individual project ideas, I.

F

E

That you should write in the same document, but maybe Orlick knows how this should be done best, but you could write like your ideas and say that, okay, if we want additional cloud storage with different providers, then we can perform. We should write this plug-in and then you write details about the plug-in you would like to additionally implement that's what I would say so I would say it should be in the same document but I'm I'm, not very strict. With this.

E

What do you think Holly yeah.

B

I think with such multiple ideas in your project proposal, so what we need is to actually understand what would be your approach. For example, how you start the exploration, how you make decisions, for example, you can say that even the community bonding and Trust building phase you increment something genetic and then you have a decent maintained point where you decide to be special, or should we proceed so as long as it's traceable as a project is perfectly fine for optional conditions in the proposal.

D

Okay, so if maybe likely, if time permits and gotten to do this sort of stuff- yes right- and it is a good part- so I really help you a lot, because I really don't know which one to I don't know how to connect concretely ideas in my mind, so it is so. If anyone has got me a lot of questions, please call I'm done with my questions.

B

You have a questions.

B

50 so my touch with the coal so yeah we just thinkin, be chatting so there's no questions probably posed on.

G

Okay, so like any.

A

G

Question if no one hello I need a bit okay, yeah, so probably I need to share my screen.

G

Okay, I hope it's visible.

B

G

So like in the previous question, one doubt was created regarding this point that better by implementing this point, can we you know discard both box position builds or we can only just discard the builds only by using the information through fingerprinting, so I think I and Martines about discussing and like we would like I was. You know a little bit confused regarding.

E

This maybe I can give a bit of context and continue from reactor so Oleg last session. We have discussed also with Wharton, and we have suggested that.

E

So external workspace manager plug-in is integrated with fingerprints. So if you get the fingerprint for specific world space, you have all the information you need to actually delete that workspace. So Martin and I suggested a to to make use of this fingerprint.

A

E

Be able to integrate with the discard plug-in and be able to delete the workspace? Does it make sense? What.

G

I'm saying yeah, it definitely made sense, but I think like wrong, so you know getting the workspace bad and all that we can also get some information about. The builds like the build information right so like what my vision is that enemies discard abusing the information or we need to go through another path for is cutting them bills, which are going one in these multiple workspaces.

E

That the thing is I, don't know how the discard plug-in works, but I would assume that you just give that plug-in a path, and you have.

G

E

G

Yes, so like I, just given over you that that delete method which is currently used by this log reader, what is a, what does it do is like it. First of all path is the path of the root directory and by getting the root directory, it uses some logic and interest discards the builds. So what I thought is that by the fingerprinting we can definitely discover places, but we can also get the path of each route directly and by using that we can.

G

You know just reach to that will and we can discard that. Will too.

E

B

Not the main purpose of this painting, yes, you can do it.

B

Only disturbing workspaces and table yields, it should be the end ho. So when something just decides to discuss the build, the process happens and then they will be stopped in bottle spaces, as we discussed before is a listener, and then you will be discussing in this pendant workspaces as a particle to discuss procedure but yeah. If you talk about specific part or external workspace manager, integration I would expect it to be managing on the workspaces, because the duration will be.

G

Okay, so like for a specific implementation with the external access manager, you suggest of discarding only the workspaces not built right now, yeah, it can be done, but at a higher level right, yes, okay, so I just include the so okay so like. According to this point, the workspaces will be discarded according to the user. So I think this point is much more clear.

G

Okay, so Martin, sir I think this comment can be resolved right.

H

Right I wanted to make sure that their understanding was was correct.

H

Workspaces do not contain bills. It's yeah.

G

H

It's build objects that have references to what spaces ultimately yeah.

G

Okay, so I, okay and the one more point which I need to discuss with you Alexa is that okay yeah. So your comment regarding to this feature, which I mentioned according to me so yeah, so actually what I meant by this is that I wanted to. You know: I had a feature like some type of recycle bin.

G

You know where, if, if due to some user mistakes, a few of the bills can discarded which were important but and now, if once they discarded and if a particular user want some information regarding that, so you know by adding this feature, it can be accessed so yeah. So definitely I have written that if time permits and if all my other features gets completed, then I will try to implement this. So you know so like here. You have just commented that it might be preferable to have it on a higher level outside this.

G

Looking so yeah yeah. Definitely it's right, but I have you know just added a few points regarding to this in the in like in this extra feature, part that maybe this can meet a you know, work for that.

C

One comment to discarding workspaces: the keyring did plug-in is using workspaces for triggering. If you have wildcards, there would not be a problem. If you discard the workspace, then they were trigger itself again.

G

Yeah so like, and that works this would be get it but like what I am stressing it like a PT like press mean, you know where, for some of us, some details of the discarded bills of workspaces can available for the users alike, in any usual case of emergency or like due to some enter like using any mistake. If anyone wants of any information regarding that, then you know we can just go where the user dead facility.

C

Actually, I don't know historically why it needs the workspace for triggering if it's wildcard it, but the problem need to be. We looked at so we don't get into a continuing, triggering deleting workspace, etc.

C

G

Yes or like I just have a look on it.

C

G

C

Current day, plugin will use the workspace for appalling, the or fetching from the remote server in order to figure out. If there is a new commit, for example, if your and then it will also trigger if it doesn't have workplace so the first time you configure, it will actually restore our workspace. So if you're then deleting it, then basically your keep infinite loop of bills.

G

Okay, yeah yeah, okay, I, just gotta, do yoga so like I, just have a look on it, one to one and oh yeah, one more thing terrified with elixir that he has committed you're dead. Why I haven't indicate a structures point in the proposal, so only so what I mind or that you know. Most of all, you know yeah so like.

B

You give me some time to process your feedback. Oh yeah, okay, sure sure blue rolling I mean because yeah I haven't taken a look at your proposal. Yet I'll try to do tomorrow on Friday.

G

Yeah so like so like I, you know just mention like there are few changes needed opposition like I. Just do it. You know by by one or two days and then I'll, you know just add a message in the Gita chatroom of that she saw bill discarded to you. Don't forgive you so at that time you can give you. You know, Bob the latest proposal of mine, mhm okay and one more doubt that.

G

Regardless of you know giving these proposals for review to team antigen-specific heater chat room where should I post my document link so that you know any mentor or of no design. You know from the community can give comments or use Asian origin.

B

And so yeah, if you want to ask something specific to development, the best way to do that is das, converging to developer memories, or if you have specific questions not related to your project directly and the results Jenkins guitar channel there is IRC channel. So you can use these generic resources on how to do ask common questions. It would be my recommendation if you want more people participating in the review.

G

Yeah, okay, sure so like I, just post my link, you know like after my final proposal, you know after your one or two reviews and after all the changes I just post the link in the or meaning list. One will be the g-shock open for all these minimis and second will be the developer of military, so that the mentors or the community you know can just go through the booze and I can get more and more reviews exactly yeah sure. Thank you.

B

So we have a few minutes left. So if there are any topics to discuss, we have several mentors on the call. Maybe we can briefly do that.

B

If it's just sitting in my little chat because as I said in the chat, I will think how to improve the communication and these Connie meetings, because here reviewing it somehow 10 box. So the questions are here. We should really do the most of technical discussions in the many lists or negative channels, because you know it may be complicated.

B

Some people, like the context.

G

The answer and relation one more because that sound like, if you you know just see that any other section needs to be a present in my proposal. Then please just inform me so that you know I can add that section like like in my proposal. I am taken care for most of the sections. You know, like my personal information and synopses project, description timelines and all that, but- and you know my working time commitments and all that but like if you you.

B

G

B

The 15 pages, we don't really want to submit proposals in several editions. Oh I think on critical sessions addressed there. If you want to add something, it would fit to do it but yeah, as I said before, usually proposals from two to five pages. So when you go beyond yeah most likely it's about additional information, you put appendix yeah.

G

Okay, yeah so like oh I'll, just go through it once and you know: I just remove the information which is not much more relevant booty so too late, so that the sighs, Emily deuce, okay.

B

So if there is no other questions, thanks, everybody for participating can again we have charts. We have other public resources and I see you that you Casey.