Handling duplicates in Dockerfiles families:
Learning from experts

Corpus

The corpus used for our experiment can be found here: corpus

All the duplicates we analyzed to identify the underlying reason for their exitence can be found here: Duplicates

The reason behind each duplicate id can be found here: Duplicates reasons

How many Dockerfiles do you manage?
This survey is for maintainers managing more than a single Dockerfile, do you still want to continue the survey?
Do your Dockerfiles share duplicate instructions (see the following screenshot)?
Have you ever updated several Dockerfiles in one commit to perform the same changes?
How do you qualify such updates (the ones that perform several Dockerfiles)?
You never updated several Dockerfiles in one commit. Can you explain why?
Would you like to have a tool to detect the duplicate instructions?
Why don't you want to have a tool to detect duplicate instuctions?
In your opinion, what should be the minimal size for a duplicate bloc of instructions to be detected?
Across how many Dockerfiles should these instructions/blocs of instructions be duplicated to be detected?
Do you have tools to avoid these duplicate instructions across your Dockerfiles?
What tools do you use?
Are you satisfied with your actual tool?
Why don't you use a tool to avoid these duplicate instructions?
Would you like to have a tool built to avoid duplicates in Dockerfiles?
Why don't you want to have a specific tool to avoid these duplicates?
Do you think the Dockerfile DSL (Domain Specific Language) should have reuse mechanisms?
What should these mechanisms help you reuse?
Why do you think the Dockerfile DSL shouldn't have reuse mechanisms?
If you want to have feedbacks about the experiment, please leave your e-mail (will be deleted when the study is over)

The results we gathered through our survey can be found here: Survey results

All variants and reuse strategies used by projects in our corpus can be found here: Reuse strategies and variants