Opened 16 months ago
Closed 6 months ago
#67823 closed defect (fixed)
R-commonmark Macports page duplicates test deps as lib deps
Reported by: | barracuda156 | Owned by: | |
---|---|---|---|
Priority: | Normal | Milestone: | |
Component: | website | Version: | 2.8.1 |
Keywords: | Cc: | ryandesign (Ryan Carsten Schmidt), arjunsalyan (Arjun Salyan) | |
Port: | R-commonmark |
Description
R-commonmark
page on Macports duplicates test deps as lib deps: https://ports.macports.org/port/R-commonmark/details
Portfile does not do that: https://github.com/macports/macports-ports/blob/master/R/R-commonmark/Portfile
They should in fact only be test deps: https://cran.r-project.org/web/packages/commonmark/index.html
Where is this coming from?
Change History (8)
comment:1 Changed 16 months ago by ryandesign (Ryan Carsten Schmidt)
Cc: | arjunsalyan added |
---|
comment:2 follow-up: 3 Changed 16 months ago by ryandesign (Ryan Carsten Schmidt)
They used to be lib dependencies before they were changed to test dependencies. Maybe the web app is forgetting to clear out previously-declared dependencies when updating a port.
comment:3 Changed 16 months ago by barracuda156
Replying to ryandesign:
They used to be lib dependencies before they were changed to test dependencies. Maybe the web app is forgetting to clear out previously-declared dependencies when updating a port.
I thought about that, but then I would have at least revbumped the port; turned out, version was updated.
comment:4 follow-up: 5 Changed 16 months ago by ryandesign (Ryan Carsten Schmidt)
The web app maintains a database with information about ports and their dependencies and other metadata. Periodically a script runs that updates that database with whatever has changed since the last time it ran. Imagine if the script runs pseudocode like this:
for each port that was changed for each type of dependency (lib, test, build, etc.) if the port has that type of dependency add or update that dependency data for that port
In other words, prior to the commit I mentioned, information about R-commonmark 1.8.0 was in the database, including that it had three lib dependencies. After that commit, the script noticed that R-commonmark was updated to 1.9.0. The script noticed that it has test dependencies so it added the test dependency information to the database. The script noticed that the port did not have lib dependencies, so it didn't do anything, erroneously leaving the previous lib dependencies in the database.
I don't know if that's how the web app's code works, but it's a guess about what could have happened, based on the apparent state of the database today from what we can see on the web site.
comment:5 Changed 16 months ago by barracuda156
Replying to ryandesign:
I don't know if that's how the web app's code works, but it's a guess about what could have happened, based on the apparent state of the database today from what we can see on the web site.
Makes sense. Well, not a big deal after all, it is just the website and the issue seems incidental.
comment:6 follow-up: 7 Changed 7 months ago by ryandesign (Ryan Carsten Schmidt)
The information on the web site seems to match the port now. Maybe the web site just hadn't completely updated before.
comment:7 Changed 7 months ago by barracuda156
Replying to ryandesign:
The information on the web site seems to match the port now. Maybe the web site just hadn't completely updated before.
Yes, can be closed.
comment:8 Changed 6 months ago by ryandesign (Ryan Carsten Schmidt)
Resolution: | → fixed |
---|---|
Status: | new → closed |
Maybe there is a bug in the way that the MacPorts web app imports its port information.