# Redesign of the \`verdi node graph generate\` command **URL:** \ **Category:** Design Discussions\ **Created:** [May 27, 2024, 10:09am UTC](https://aiida.discourse.group/t/redesign-of-the-verdi-node-graph-generate-command/395 "2024-05-27T10:09:19Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![sphuber](https://yyz2.discourse-cdn.com/free1/user_avatar/aiida.discourse.group/sphuber/32/6_2.png) [@sphuber](https://aiida.discourse.group/u/sphuber)\ **Post date:** [May 27, 2024, 10:09am UTC](https://aiida.discourse.group/t/redesign-of-the-verdi-node-graph-generate-command/395/1 "2024-05-27T10:09:19Z")
The current interface is: ```auto verdi node graph generate ROOT_NODE [OUTPUT_FILE] ``` That is, currently, only a single root node can be specified. But it is perfectly reasonable to want to specify multiple root nodes. To this end, [this pull request](https://github.com/aiidateam/aiida-core/pull/6405) was submitted and merged. It adds the `-N/--nodes` option that now allows users to specify multiple root nodes. The original `ROOT_NODE` argument was deprecated. But maybe an even better design would be to change the interface to ```auto verdi node graph generate ROOT_NODES ``` That is, allow for one or multiple root nodes as positional arguments and the output file can be specified through an option (e.g., `--output-file`). The advantage is that this is more in line with other commands that take multiple nodes as input. They are almost always specified as arguments, with the notable exception being `verdi archive create`. The only downside is that this would be breaking backwards compatibility, whereas the current solution merely deprecates the old behavior. So, the question is, **which solution do you find more intuitive** : ### Option-based ```auto verdi node graph generate 100 verdi node graph generate 100 output.aiida verdi node graph generate -N 100 101 102 verdi node graph generate -N 100 101 102 -- output.aiida ``` ### Argument-based ```auto verdi node graph generate 100 verdi node graph generate 100 --output output.aiida verdi node graph generate 100 101 102 verdi node graph generate 100 101 102 --output output.aiida ``` Even though the argument-based approach would be breaking some use cases, I think it is the better solution and would vote for adopting that over the current option-based approach. **Please voice your opinions below on what you think is the best solution.** ---
**Author:** ![ali-khosravi](https://yyz2.discourse-cdn.com/free1/user_avatar/aiida.discourse.group/ali-khosravi/32/58_2.png) [@ali-khosravi](https://aiida.discourse.group/u/ali-khosravi)\ **Post date:** [May 27, 2024, 12:30pm UTC](https://aiida.discourse.group/t/redesign-of-the-verdi-node-graph-generate-command/395/2 "2024-05-27T12:30:15Z")
Thank you for providing a compact and well-summarized question. With the current implementation of `verdi` line, to me argument based is more intuitive. A side node, I feel it would also be even more user friendly to accept ranges: ```auto verdi node graph generate 100 101 102 --output output.aiida verdi node graph generate 100-102 --output output.aiida ``` ---
**Author:** ![mbercx](https://yyz2.discourse-cdn.com/free1/user_avatar/aiida.discourse.group/mbercx/32/345_2.png) [@mbercx](https://aiida.discourse.group/u/mbercx)\ **Post date:** [May 28, 2024, 11:48am UTC](https://aiida.discourse.group/t/redesign-of-the-verdi-node-graph-generate-command/395/3 "2024-05-28T11:48:08Z")
As already discussed yesterday, I’m also in favor of the argument-based approach. 🙂 The node ranges idea is also interesting, but I would never dare to introduce such [scope creep](https://en.wikipedia.org/wiki/Scope_creep) to this discussion. 😇 ---
**Author:** ![danielhollas](https://yyz2.discourse-cdn.com/free1/user_avatar/aiida.discourse.group/danielhollas/32/4_2.png) [@danielhollas](https://aiida.discourse.group/u/danielhollas)\ **Post date:** [May 28, 2024, 3:46pm UTC](https://aiida.discourse.group/t/redesign-of-the-verdi-node-graph-generate-command/395/4 "2024-05-28T15:46:29Z")
Given that the OUTPUT\_FILE is optional, it doesn’t really make sense for it to be a positional argument. So I am also in favour of the argument-based approach. > [@ali-khosravi](#): > > A side node, I feel it would also be even more user friendly to accept ranges: I don’t think the extra complexity is worth it in this case — supposedly the case where root nodes would have consecutive IDs should be very rare. ---
**Author:** ![sphuber](https://yyz2.discourse-cdn.com/free1/user_avatar/aiida.discourse.group/sphuber/32/6_2.png) [@sphuber](https://aiida.discourse.group/u/sphuber)\ **Post date:** [May 28, 2024, 9:02pm UTC](https://aiida.discourse.group/t/redesign-of-the-verdi-node-graph-generate-command/395/5 "2024-05-28T21:02:55Z")
Thanks everyone for the feedback. Seems everyone was on board with breaking backwards compatibility in order to have the most intuitive interface: [CLI: `verdi node graph generate` root nodes as arguments by sphuber · Pull Request #6427 · aiidateam/aiida-core · GitHub](https://github.com/aiidateam/aiida-core/pull/6427) ---
**Author:** ![geiger\_j](https://yyz2.discourse-cdn.com/free1/user_avatar/aiida.discourse.group/geiger_j/32/57_2.png) [@geiger\_j](https://aiida.discourse.group/u/geiger_j)\ **Post date:** [May 29, 2024, 7:14am UTC](https://aiida.discourse.group/t/redesign-of-the-verdi-node-graph-generate-command/395/6 "2024-05-29T07:14:31Z")
A bit late to the party, but agreeing with the rest of the team. Just ran the command with multiple nodes, and, intuitively, I was expecting multiple output files, one for each graph, rather than a single one. But I guess if a user wants this behavior, they can just loop over the nodes outside the `verdi` command, so all good!