can someone build a GUI for this great cli?

[Chesterguan]

People who play with CSVs daily mostly do not know how to program. Otherwise, Python can solve most jobs.

Rust is fast, but do you guys have a plan to build the GUI for it? Or just leave the CLI for AI agents?

[Yomguithereal]

People who play with CSVs daily mostly do not know how to program.

Why would you assume that? People who deal with nontrivial CSV data (typically data that you cannot open in spreadsheet software like Excel or LibreOffice) usually know programming and how to use CLI tools because this is the only way to deal with such data, as GUI tools will always fail them since they usually scale pretty badly.

Now regarding a GUI, we are not opposed to it per se, but we don't really know how this could work. GUIfying a CLI tool that is leveraged to deal with very large streams of data does not usually work very well. This is one of the reasons why people still use the command line nowadays, because it is the best (or at least not the worst) interface to solve this kind of problem, along with script programming.

So creating a GUI would require quite a lot of design work in the case of xan and I see two different paths there:

1. either you need to build a GUI around very streamlined use-cases that you identified to be pivotal for your users, but then you might create a GUI leveraging xan to work, and don't need to GUIfy xan itself
2. either you need to build some kind of visual programming interface so people may build complex pipelines with the tool, but in this case, you will end up re-creating a visual shell of some kind and it might be easier to teach people the command line directly (the command line looks scary to people but frankly it is not that complicated to teach)

TL;DR: we are not opposed to the idea of a GUI but we don't know how to design it to be useful and to whom it would be useful beyond the command line interface.

[Chesterguan]

@Yomguithereal Thanks so much for your response!

Yeah, I agree with most of your statement. Nontrivial CSVs surely require better tools to process, and obviously, Xan is the better tool than anything else I saw.

I am a data scientist who has been working with large medical data for years. For the complex tasks or ETLs, we surely have our solutions mostly in Python. But to be honest, the data format will be parquet, which is more convenient to play with. You are right, besides parquet, we also have tons of CSV files. From our recent investigation, lab-wise. The most struggling group is not the data team, who have a background in programming and the intention to learn to use CLI or other alternative solutions. But the clinical team, who are not good at either coding tools or even Excel/spreadsheet. That's my initial idea to wrap the Xan into a better tool for the clinical team.

appreciate for your hard work. I get your ideas, and I will try to implement or test the wrapping module in our settings. Love to have further discussion if you are available.

[Yomguithereal]

From our recent investigation, lab-wise. The most struggling group is not the data team, who have a background in programming and the intention to learn to use CLI or other alternative solutions. But the clinical team, who are not good at either coding tools or even Excel/spreadsheet. That's my initial idea to wrap the Xan into a better tool for the clinical team.

I have encountered this kind of situation in the past and the solution was usually to design dedicated tool for those publics around the data itself (we used to call those tools "datascapes" in our lab, think very customized dashboard, design to answer specific research questions), rather than to create GUI version of generic tools wrapping the stuff that yourself or myself would use, typically because acculturation to the "data" mindset is actually very close to pushing those people to learn how to code or use the command line which is interesting per se but they may not want to learn it or they don't have the time to do so.

But this is an interesting topic in itself. For instance, I have another CLI tool written in python to collect data on the web called minet. This one I would know very well how to desing a GUI around it to empower people not wanting to invest in learning the command line. But for xan itself, I find the task way harder as its purpose is very tied to manipulating data itself and does not offer a means to a particular end.