- Mar 20, 2023
-
-
Evangelos Foutras authored
Switch from xtarfile to libarchive See merge request !1
-
- Mar 19, 2023
-
-
Evangelos Foutras authored
The former was used for .zst support. It does it by decompressing the whole tarball into a temporary location. With intel-oneapi-basekit exceeding 16G uncompressed size, we're running out of space on gemini. Switch to libarchive which should be able to stop after reading .PKGINFO without decompressing and/or reading the rest of the .zst package file.
-
- Jan 15, 2020
-
-
Jelle van der Waa authored
Support .zst files by using xtarfile
-
- Jan 01, 2020
-
-
Jelle van der Waa authored
xtarfile can handle opening .zst files which sadly Python does not support yet. The xtarfile support however does not support the '|*' flag nor an encoding flag which seem to be unrequired for parsing the PKGFINO contents. https://bugs.python.org/issue37095
-
- May 25, 2019
-
-
Florian Pritz authored
Some (meta) packages like kdeaccessibility-meta do not have a license set and right now this triggers an exception. Fix it by just setting an empty value like archweb does. Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
For a KeyError, print(e) just prints the key name that wasn't found. That's not even remotely helpful/obvious. Now we get a stacktrace and the usual "KeyError: foo" message, which is much better and actually what I wanted when I changed this. Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Apparently some package doesn't have the pkginfo['license'] value set so we get an exception, but that exception isn't expected and thus we don't get the name of the broken package. Also, we really want that the uploader tries to upload all the other files instead of aborting early. We still want it to exit non-zero though. Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- May 02, 2019
-
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- Apr 28, 2019
-
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Having the default parameter in the parameter list always initializes the class. That's not what we want here since it actually create a real database when running the tests. Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- Apr 06, 2019
-
-
Florian Pritz authored
Oops. There should have been an index there... Having it should improve performance somewhat. Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- Apr 05, 2019
-
-
Florian Pritz authored
We sometimes get various types of exception (timeouts, SSL errors, broken remove servers). Running the script again will error at the same point. Instead of terminating the current run, we now try to complete as much as possible by pushing on. Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- Apr 03, 2019
-
-
Florian Pritz authored
Some pacakges have very many files (stack has ~1376) that are currently uploaded all at once. If there is an exception during this run or if the program is killed (happened due to 'open files' limit), the sucessfully uploaded files are not added to the sqlite db. Next time they are uploaded again and the upload may be interrupted again, leading to huge time loss. To fix this, we now upload in chunks of 20 files at a time and add them to the db after this. Thus, as long as we manage to upload 20 files without crashing, the next time, there will be fewer files to upload. Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- Mar 30, 2019
-
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- Mar 29, 2019
-
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- Mar 28, 2019
-
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
- Mar 27, 2019
-
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-
Florian Pritz authored
Signed-off-by:
Florian Pritz <bluewind@xinu.at>
-