Feature #9499
closedSwitch to new Clade API
0%
Description
Core components actively use Clade API that changed quite considerably recently. We need to switch to the new Clade API.
Updated by Evgeny Novikov over 5 years ago
- CI detected that there are many failures for job "Testing decomposition strategies". Most likely in some specific places Klever still does not use the new Clade API properly.
- A more generic issue is that we still have not a good approach for installing appropriate versions of Clade from PyPI. BTW, there is the same issue with BenchExec.
Updated by Evgeny Novikov over 5 years ago
One more issue that should be fixed before switching to the new Clade API: https://github.com/17451k/clade/issues/61.
Updated by Evgeny Novikov over 5 years ago
Evgeny Novikov wrote:
This feature was implemented in branch clade-interface. At the moment there are 2 issues:
- CI detected that there are many failures for job "Testing decomposition strategies". Most likely in some specific places Klever still does not use the new Clade API properly.
The corresponding issue is https://github.com/17451k/clade/issues/66.
Updated by Evgeny Novikov over 5 years ago
Evgeny Novikov wrote:
Evgeny Novikov wrote:
This feature was implemented in branch clade-interface. At the moment there are 2 issues:
- CI detected that there are many failures for job "Testing decomposition strategies". Most likely in some specific places Klever still does not use the new Clade API properly.
The corresponding issue is https://github.com/17451k/clade/issues/66.
I fixed the issue, so, waits for an appropriate decision of a problem with Clade versions and a fix of https://github.com/17451k/clade/issues/61.
Updated by Evgeny Novikov over 5 years ago
In general, almost everything works at the moment. But I expect several related changes that can influence the Clade API. So, before switching to the new API, let's wait for this API to be rather stable. In particular, this assumes that there will be a new major version of Clade that we will be able to point out in our deployment configurations.
Updated by Evgeny Novikov over 5 years ago
With https://github.com/17451k/clade/issues/75 we should change the way to store and to retrieve so called working source trees.
Updated by Evgeny Novikov over 5 years ago
Evgeny Novikov wrote:
With https://github.com/17451k/clade/issues/75 we should change the way to store and to retrieve so called working source trees.
https://github.com/17451k/clade/issues/83 is the right issue for that.
Updated by Evgeny Novikov over 5 years ago
Besides, we should check whether different variants of coverage work as before.
Updated by Evgeny Novikov over 5 years ago
Evgeny Novikov wrote:
Besides, we should check whether different variants of coverage work as before.
I don't see any issues related with code coverage under normal conditions. Let's fix particular bugs if so.
Updated by Evgeny Novikov over 5 years ago
- Status changed from New to Resolved
The branch is ready for merge. I am going to initiate its thorough testing.
Updated by Evgeny Novikov over 5 years ago
- Status changed from Resolved to Open
Numerous tests failed. But hopefully the reason seems to be the same.
Updated by Evgeny Novikov over 5 years ago
- Status changed from Open to Closed
After updating Clade from 3.0 to 3.0.1 all tests passed. So, I merged the branch to master in 583af12. Enjoy new Clade that includes great improvements.
There can be some troubles with updating existing instances of Klever. You should manually uninstall Python3 package ldv-clade before updating to the latest master. If you will forget about this, don't worry, you will be able to do this later, but also you may need to uninstall/install clade==3.0.1 manually.
Moreover, I will delete ldv-clade from PyPI to avoid any ambiguities.