Let's say there is an audio based on the http kind. The file it is downloading is very big, so downloading takes a long time.
Since this is problematic for larigira, a sysadmin might want to execute the action with large advance (manually, using cron), running this
larigira should be able to behave appropriately, and use this cached file. This doesn't work, because the http audiogenerator uses mkstemp, which creates a random file name, so this could never possibly work.
what should happen, instead?
running it twice should be idempotent.
describe all the steps we can do to reproduce the bug
run the aforementioned command two times. the second time you run it, network transfer should be very small. The same filename should be used.
## Is it a bug or a feature request?
A feature request
## Describe what happens
Let's say there is an audio based on the `http` kind. The file it is downloading is very big, so downloading takes a long time.
Since this is problematic for larigira, a sysadmin might want to execute the action with large advance (manually, using `cron`), running this
```
/usr/local/bin/larigira-dbmanage getaction 1234 | env TMPDIR=/tmp/larigira.1000/ /usr/local/bin/larigira-audiogen -
```
`larigira` should be able to behave appropriately, and use this cached file. This doesn't work, because the `http` audiogenerator uses `mkstemp`, which creates a _random_ file name, so this could never possibly work.
### what should happen, instead?
running it twice should be idempotent.
### describe all the steps we can do to reproduce the bug
run the aforementioned command two times. the second time you run it, network transfer should be very small. The same filename should be used.
Is it a bug or a feature request?
A feature request
Describe what happens
Let's say there is an audio based on the
httpkind. The file it is downloading is very big, so downloading takes a long time. Since this is problematic for larigira, a sysadmin might want to execute the action with large advance (manually, usingcron), running thislarigirashould be able to behave appropriately, and use this cached file. This doesn't work, because thehttpaudiogenerator usesmkstemp, which creates a random file name, so this could never possibly work.what should happen, instead?
running it twice should be idempotent.
describe all the steps we can do to reproduce the bug
run the aforementioned command two times. the second time you run it, network transfer should be very small. The same filename should be used.
branch
http-improveseems to be working, but need real-life tests before getting merged.