Closed prjemian closed 3 years ago
Test conda solver time (with all packages in local cache from previous install runs). Selective removal of packages from environment installation. Test using this command:
date; conda env create --force -f ./environment_2021_2.yml; date
test condition | time to complete | seconds |
---|---|---|
all packages included | 3m 5s | 185s |
only # - aps-dm-api |
2m 52s | 172s |
only # - hklpy>=0.3.16 |
1m 20s | 80s |
only # - nexpy |
3m 7s | 187s |
only # - pandas |
3 m 21s | 201s |
only # - pydm |
3m 11s | 191s |
only # - pymca |
3m 3s | 183s |
only # - xrayutilities |
3m 25s | 205s |
only # - hklpy>=0.3.16 (test again) |
1m 19s | 79s |
add- happi |
2m 54s | 174s |
Inclusion of the hklpy package in the environment installation doubles the time needed by the conda env install
process to install the environment. This result should be pursued by the hklpy package.
Add bluesky-queueserver and webclient to the environment.
Saul:
Have not yet looked at the details of implementing the bluesky-queueserver for APS. These items to be implemented (as far as I know now):
I can install the Python stack easily, the other steps will take experimentation to gain experience.
I've added Garrett Bischof (NSLS-II) to this thread. He is experienced with this software and might have a suggestion to help us forward at the APS. This web-interface to Bluesky is exciting for many APS beam lines. The recent talk at the 5LS meeting has generated much interest here.
Pete
On 6/2/2021 9:44 AM, Lapidus, Saul H. wrote:
I emailed Pete the other week about this but Im really excited about the idea of implementing the process queue server (that they showed at the 5LS meeting from NSLS 2) at 11-BM. What do you think is needed to move forward on this?
Saul
@mdwyman , @jkirchman : Any chance to examine? I'd like to merge by Friday.
Yikes! widgets and queueserver are on PyPI but not webclient
Needs a pip install from github for bluesky-webclient
The install worked except bluesky-webclient installed as UNKNOWN.
bluesky-webclient is a javascript package that uses the npm package manager. Don't put it in the conda environment.
name: bluesky_2021_2
As part of the process, identify which package(s) make the
conda env create
process take so much longer to solve.