Closed jazdrv closed 6 years ago
It's going to take me a long time before I can much comment on the sort algorithms. I only have a few observations so far
There are no comments. I'm no model for great comments, but I can see the impact of having none, when trying to understand what the code is supposed to be doing
getVCFvariants should be part of the data load work package. It's probably not quite correct yet, but I don't see it as part of sort
genotype in vcfcalls, as we've discussed, is not desirable. Parsing VCF files and storing the data is one thing, and analysis is another thing. The interface for genotype is get_variant_array or thereabouts
there should not be a db1.py. Make the changes you need in db.py
I have gotten further with this review but still have more to go. Here are some things so far.
Approach to getting this merged with master is murky. I have previously suggested a strategy, and I'll reiterate it here. In slack, Feb 20, in #coding-database, Zak asked about the approving pull requests. I answered. I thought that was clear now, if unfamiliar. "this process is new to me, I'm afraid. so silly question ... the idea here is that I just review the code changes in the browser? [do I approve it?]" "yes. I don't think we've really formalized any process (and that's OK with me at least for now) but the idea is changes get reviewed by someone before going into master and ideally any issues identified in the review get discussed/resolved before it gets delivered to master"
Because there are changes to files like lib.py and db.py versus latest on master, I don't think I'm really reviewing the correct thing. Until it's in a pull request or diff off of master latest, I can't be very precise with my review comments.
March 13, I suggested a strategy: "What I think needs to happen is Zak prepare incremental branches for delivery to master that he can do a pull request on and we can review properly."
I recommend breaking this up into incremental deliveries. It can be done as one large delivery, but I don't recommend that because it's quite a bit more difficult to address review comments and make sure the code is all caught up. It's also harder for the reviewer. This is just basic, not specific to this project/branch. There's more than one way to accomplish this, but and here's one possible workflow:
So far, here are some review comments:
I'll probably have more comments later. Resolving these in an issue is not the way I'd prefer. I'd prefer to resolve review comments via a pull request.
Hi Jef, Zak, I think most of Jef's comments are already acknowledged, particularly #1, #3, #4 and #5, particularly since Zak has been working on this as a "proto-type", so the integration and quality of the code presumably isn't up to as high a quality as would otherwise be the case. Nevertheless, the outputs do show that this is roughly the logic principles that we need. I'd prefer to see what Zak's done as a measure that lets us prove what we've got there is viable, but I think we all know it will take some migration to get it properly integrated. My question you both is whether you think you can do this together, let's say over the next two weeks, with minimal input from me? Zak: are you able to create one or more intermediate branch points as Jef suggests? Jef: are you prepared to take charge of organising and performing this merger? I'll address Zak's e-mail separately, because it deserves a detailed response. Cheers, Iain.
From: jeftreece <notifications@github.com>
To: jazdrv/dnaTools dnaTools@noreply.github.com Cc: moregubbins gubbins@talk21.com; Assign assign@noreply.github.com Sent: Thursday, 22 March 2018, 0:16 Subject: Re: [jazdrv/dnaTools] sort code needs code review (#19)
I have gotten further with this review but still have more to go. Here are some things so far.Approach to getting this merged with master is murky. I have previously suggested a strategy, and I'll reiterate it here. In slack, Feb 20, in #coding-database, Zak asked about the approving pull requests. I answered. I thought that was clear now, if unfamiliar. "this process is new to me, I'm afraid. so silly question ... the idea here is that I just review the code changes in the browser? [do I approve it?]" "yes. I don't think we've really formalized any process (and that's OK with me at least for now) but the idea is changes get reviewed by someone before going into master and ideally any issues identified in the review get discussed/resolved before it gets delivered to master"Because there are changes to files like lib.py and db.py versus latest on master, I don't think I'm really reviewing the correct thing. Until it's in a pull request or diff off of master latest, I can't be very precise with my review comments.March 13, I suggested a strategy: "What I think needs to happen is Zak prepare incremental branches for delivery to master that he can do a pull request on and we can review properly."I recommend breaking this up into incremental deliveries. It can be done as one large delivery, but I don't recommend that because it's quite a bit more difficult to address review comments and make sure the code is all caught up. It's also harder for the reviewer. This is just basic, not specific to this project/branch. There's more than one way to accomplish this, but and here's one possible workflow:
I'm not sure what might still be open here, but I'm closing this issue with merge #38
concerns: