-
Notifications
You must be signed in to change notification settings - Fork 24
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Some pages show a table of packages for an expired flavor/flavour #449
Comments
I need a summary aimed at management please. |
Please, I really need you to tell me what you think the problem is. Please put that as the first sentence. It really helps me. In this case I think it is: This port has 3 flavors listed. The list of packges includes flavors not listed. Current status: stale data being kept for repos not recently updated. |
I'm not sure of problem, because I have not paid much attention, in the past, to flavours and their representations. Above, I added a line:
HTH |
re: https://github.com/FreshPorts/freshports/blob/main/classes/package_flavors.php#L35
|
Some stuff from debug output:
|
I wonder if the issue is stale data. Not everything is clear out of the database when reading in a new set of packages. |
|
I think this is a back-end issue in https://github.com/FreshPorts/packages-import |
There is a delete done: https://github.com/FreshPorts/packages-import/blob/master/import_packagesite.py#L56 Let's look the |
What do we fetch? https://github.com/FreshPorts/packages-import/blob/master/fetch-extract-parse-import-one-abi.sh#LL48C15-L48C72 says:
|
|
There it is. It's
|
And the other one is there too.
|
Conclusion:
Action required:
|
Here are the
|
And the
|
I think this |
Here we go. It's the slave ports.
|
So why does FreshPorts list I see the
|
The data gets broken up like this:
|
Doing that manually:
This shows why Why does it also get listed under This is what I find for
|
Next, look at the data import code. |
https://github.com/FreshPorts/packages-import/blob/master/import-via-copy-packagesite-all-raw-fields.py#LL52C26-L52C38 imports the file ( I think this is the |
|
Here's the data now in that table. It seems right.
|
This is the table which gets populated from
|
This content also looks good.
|
Going to the website code (https://github.com/FreshPorts/freshports/blob/main/classes/packages.php#LL51C25-L51C37), and mangling the
|
I'll come back to this. I want to look at how the package details are composed onto the page. |
Past 02:10 in the early hours, I need to sleep, but first,
Better to my sleepy non-developer eye:
From #449 (comment) regarding
A tentative conclusion, yes, partly because of that. Also, partly because of |
Now I'm not sure why |
The data is in the database. This is not a caching issue.
|
The data is in the database. It's not a website display or caching problem. Let's look at package processing. ... which we already did at #449 (comment) |
|
No, that just deletes from The next place to look: the population of That code invokes updatepackagesfromrawpackages |
Looking at that code (which is not on GitHub), it does this
|
Next for me:
|
Reading https://news.freshports.org/2020/04/17/1325/ I found confirmation that a @grahamperrin: The root cause has been found.
|
First thought: Go with delete. That will make Second thought: We could stay with option 1, and then just Third: wait, we already add |
However Why would it be just one ABI? If we import an ABI today, and never again import that ABI? Those values will remain in the table forever. Let's look here:
That's all FreeBSD 12 - not sure if that's relevant. |
Let's use https://test.freshports.org/archivers/py-lz4/#packages as an example. It shows 5 flavors:
However, we have only this:
But
Let's delete No,that didn't affect the page. However, deleting from |
Now my theory is: stuff hanging around in The Why not wipe |
The test plan:
Hypothesis After an import of any ABI/package_set combination the following will be true:
EDIT: I said |
|
After clearing cache, https://test.freshports.org/archivers/py-lz4/#packages shows only three flavor:
|
Next: invoke This looks like no updates at all.
|
Well, we still have Where did that come from. |
Well, we have this:
|
I understand now. The problem is not the delete statement. It is correct. The problem is old For example, the last time The packages are still out there. |
This is resolved if FreshPorts deletes Does an authoritative list exist of ABI for which packages are built? |
I guess, the FreeBSD Packages Management Team https://www.freebsd.org/administration/#t-pkgmgr will know. HTH |
email sent. |
Resent on Oct 15th. Sent to |
Might be related to: |
@grahamperrin this seems fixed/ready to close? |
devel/git
Three flavours listed, five tables of packages, two of the tables appear to be for slaves that are not flavours.
https://www.freshports.org/devel/git/#flavors three flavours (git, git-lite, git-tiny).
https://www.freshports.org/devel/git/#packages five tables, I'm not sure whether these two (of four) slave ports belong:
Slave ports that are not on the same page:
devel/git-gui
https://www.freshports.org/devel/git-gui/ version 2.38.1_4
https://www.freshports.org/devel/git-gui/#packages 2.40.1, in some cells of the table, exceeds 2.38.1_4 …
devel/git-svn
https://www.freshports.org/devel/git-svn/ 2.38.1_4
https://www.freshports.org/devel/git-svn/#packages 2.40.1, in some cells of the table, exceeds 2.38.1_4 …
https://www.freshports.org/devel/git-svn/#history 2.32.0_1 is inferior to 2.38.1_4.
Whilst drafting this, I remembered:
Is this report entirely a duplicate of 435?
The text was updated successfully, but these errors were encountered: