Home → Questions →

Making tests a prerequisite for others; Caching in between these tests

• Edit

noel.kuntze

15 Mar, 2020 05:39 PM

Hello,

Is it possible to make groups of tests, make some depend on others and execute them in order? I'd like to build some dependencies first, cache them and make that cache available to the jobs that depend on them. That would significantly decrease build time for me because large libraries would only need to be built one time instead of N times. I guess that would also require some change to the cache semantics.

Kind regards

Noel

1. Support Staff 1 Posted by Feodor Fitsner on 17 Mar, 2020 07:26 PM

   

   Hi Noel,

   Do you mean dependencies between build jobs within a single build?

   • Edit

2. 2 Posted by noel.kuntze on 18 Mar, 2020 12:14 PM

   

   Hello Feodor,

   Yes, exactly.

   • Edit

3. Support Staff 3 Posted by Feodor Fitsner on 18 Mar, 2020 08:01 PM

   

   Yes, it's possible. Look at this yaml template defining 3 jobs running in a sequence:

   environment:    
     matrix:
   
   # Hyper-V
   # =======
   
     - job_name: Part 1
   
     - job_name: Delay between parts
       job_depends_on: Part 1
   
     - job_name: Part 2
       job_depends_on: Delay between parts
   
   matrix:
     fast_finish: true
   
   init:
     - appveyor version
     - echo Some common logic here...
   
   test: off
   
   for:
   
   # Part 1
     -
       matrix:
         only:
           - job_name: Part 1
   
       build_script:
       - echo Part 1
   
     -
       matrix:
         only:
           - job_name: Delay between parts
   
       build_script:
       - ps: Start-Sleep -s 60
   
   # Part 2
     -
       matrix:
         only:
           - job_name: Part 2
   
       build_script:
       - echo Part 2
   

   • Edit

4. Support Staff 4 Posted by Feodor Fitsner on 18 Mar, 2020 08:03 PM

   

   You can also have groups of jobs and then depend on a group, for example:

   environment:    
     matrix:
   
   # Hyper-V
   # =======
   
     - job_name: Part 1 - A
       job_group: Part 1
   
     - job_name: Part 1 - B
       job_group: Part 1
   
     - job_name: Delay between parts
       job_depends_on: Part 1
   
     - job_name: Part 2
       job_depends_on: Delay between parts
   
   matrix:
     fast_finish: true
   
   ...
   

   Now jobs Part 1 - A and Part 1 - B will be running in parallel and Delay between parts will run when all jobs within Part 1 group are finished.

   We are going to document this functionality this week.

   • Edit

5. 5 Posted by noel.kuntze on 19 Mar, 2020 10:09 AM

   

   Hello Feodor,

    

   Thank you for your insight! I will test this out later. :)

    

   Kind regards

    

   Noel

   • Edit

6. 6 Posted by noel.kuntze on 27 Mar, 2020 09:48 AM

   

   Hello Feodor,

   How are the cache semantics of appveyor exactly? Is the cache updated in between jobs already? So can i pass data from job 1 to job 2 via a cache, if they run consecutively?

   Kind regards

   Noel

   • Edit

7. Support Staff 7 Posted by Feodor Fitsner on 01 Apr, 2020 05:31 PM

   

   Hi Noel,

   The cache is job-specific. However, you can pack your dependencies (intermediate results), upload (deploy) to Azure storage, S3 or FTP account and then download/unpack in the next jobs. This is what, actually, AppVeyor cache does.

   • Edit

8. 8 Posted by noel.kuntze on 02 Apr, 2020 02:16 PM

   

   Hi Feodor,

   Thank you for your response.

   Kind regards

   Noel

   • Edit

9. noel.kuntze closed this discussion on 02 Apr, 2020 02:16 PM.