Package distribution method for RISC-V Sail model #39
Comments
|
I am interested in accepting this task, and I think PLCT Lab's technical background can help. Since this week and next week are the Chinese New Year, I estimate that I can return some research results and plans around February 28 at the earliest. |
|
@lazyparser and PLCT have expressed interest. I've also asked Greg Sterling to engage here too. I'll get him added to the issue as soon as I can. |
|
Fedora has already built all the dependencies and tested |
|
Thanks for the information. @gsterlin, please take a look at this. |
|
There are Debian packages about sail/sail-riscv and their dependent packages. Meanwhile, ocaml-linenoise, omd and lem has been pushed Debian salsa repo under Debian OCaml team and was waiting to sponsor them. Other packages are doing similar things. |
|
Given how rapidly Sail is being modified at the moment to support new features for the model I’m not convinced having packages in Debian stable is at all useful. It seems highly likely that by the time they are released they won’t be usable. |
|
Make sense. If we want to use a fresh deb/rpm package, we would better to keep one final package, in other words, to keep sail-riscv package with an upstream distribution. But before this, these dependent packages should be packaged into distribution to reduce the maintenance burden. |
|
My belief is that even though we may be iterating over packages faster than we'd like, it would also give us some ability to talk across differing distributions to discuss what versions of software, and specifically which packages were used, that gave specific results. This should help us consider branching strategies for the models, and being able to provide specifications about what went into the model. My hope is that this would give us a more common framework from which we could discuss changes, and test them out in more consistent ways. It is also possible that I need to spend more time understanding how things work and come together for sail, and I am open to learning more about how things work today, especially where we see issues or possible room for improvement. |
|
bisect-ppx has been packaged to the Debian OCaml namespace and uploaded by a Debian Developer. |
|
sail-riscv on Fedora docker image has been built. |
|
@gsterlin, would you kindly collect all the progress and summarize this? I suspect that there's benefit in knowing what pkgs are in which ecosystem. Then, we can tie this into the whole broader plan. Thx. |
|
https://github.com/orgs/fedora-sail/repositories The spec files for lem linksem ocaml-omd ocaml-linenoise ocaml-pprint have been pushed |
|
On Debian side update: So, until now. bisect-ppx has been uploaded to Debian NEW queue and ocaml-linenoise are already in Debian official. We still have omd, lem and linksem need to be uploaded to Debian for sail/riscv-sail. |
|
Debian： Still waiting for omd to be reviewed/uploaded. No progress on the other packages now. |
|
Fedora package still waiting for review…… |
|
ocaml-omd ocaml-linenoise ocaml-pprint |
|
Debian: lem has been uploaded to FTP NEW queue. |
|
ocaml-pprint has been reviewed and approved |
|
@gsterlin, would you mind sharing your progress for the next meeting on August 6? |
|
Absolutely! A quick note is that I have been working with Carl to leverage some tools (asdf/mise, and direnv) to do the environment variable/configuration and tool version management. This replaces the code and configuration files I had written, greatly simplifying the problem. We have made a lot of progress over the last two weeks and will have some content to share shortly that will help new comers work with our environments, and provide the building blocks for enabling discussions about which versions of packages or environment variables are necessary for proper environments per repository. |
|
Debian packaging: FTP master has commented on lem and waiting for review again. |
|
ocaml-omd was also approved! |
|
Debian packaging: no update in the past two weeks, still waiting for lem to be approved to upload. |
|
fedora: No updates in the past two weeks |
|
@gsterlin, I wonder if Wei Fu can provide some guidance on the ocaml-linenoise PR. |
|
Wei Fu is my mentor, but I think we should contact the maintainer of ocaml-linenoise (or Fedora packager) |
|
Sorry, my English may cause some misunderstand, let me clarify |
|
Since there has been no response from the upstream of linenoise, I sent the patch to the downstream of Fedora, but I think Fedora may not accept this patch (because Fedora requires the upstream to accept the patch first). |
|
Good news! Fedora, as the downstream, has already merged this change! |
|
Debian packaging: linksem is packaging and it is waiting for lem's one commit. Summary: Now we only have linksem and sail left as dependencies of riscv-sail that need to be uploaded to Debian. |
|
Since the linenoise patch was merged by Fedora, ocaml-linenoise is being reviewed again |
|
Debian packaging: linksem is being reviewed by my sponsor. |
|
The reviewer has not responded yet |
|
There is a small change in the linenoise package on fedora. Due to my mistake in the past, the symbols were not exported. Fortunately, the linenoise package on fedora has been merged ocaml-linenoise is waiting for review again |
|
Debian packaging update: |
|
Thanks, folks! |
|
No update on Debian packaging in past two weeks. |
|
ocaml-linenoise passed, no problem now! Now wait for linenoise to be rebuilt |
Technical Group
Golden Model SIG
ratification-pkg
Technical Debt
Technical Liaison
Bill McSpadden
Task Category
SAIL model
Task Sub Category
Ratification Target
3Q2023
Statement of Work (SOW)
Component names:
Sail
Requirements:
Currently, the RISC-V Sail model must be built (compiled) by the user. It has been requested that the model be compiled and distributed using industry-standard package management tools.
Deliverables:
Acceptance Criteria:
Projected timeframe: (best guess date)
6 person months
SOW Signoffs: (delete those not needed)
Waiver
Pull Request Details
No response
The text was updated successfully, but these errors were encountered: