[rpdillon]

This is a particular brand of take strikes me as lazy. In general, each type of product is going to have some core features that almost everyone needs. And then there's going to be a long tail of features that fewer and fewer users need to make use of the tool effectively. Office tools like LibreOffice and Google Sheets strike a sort of 80/20, where they can build perhaps less than half the features of the totally complete product, but still serve a huge percentage of the market's needs (maybe 95%+, since most users aren't power users).

So when I see critiques of GIMP versus Photoshop, or Linux versus Windows, or LibreOffice versus Microsoft Office, saying "oh, it has fewer features and therefore nobody can take it seriously" it's just reductive, and provides zero useful insight. It's all about the particular needs of the person or organization and how those intersect with the features of the product they're thinking of adopting.

[yummypaint]

I would go further and say that MS products have completely backwards priorities that lead to an overinflated feature count. What good is fancy formatting in excel if it chokes and crashes once the file hits about 20 MB? Yet despite all this emphasis on form over function over multiple decades of being a flagship product for a multibillion dollar software empire, it still produces plots that are unacceptable for publication and instill bad habits in students.

I'm convinced the people who insist on "features" in these products don't actually use them, because if they did they would realize they suck and are a distraction from a poor core product. It's like people in the US who live in downtown apartments and insist on driving massive overpriced pickup trucks to commute to work and get groceries, never hauling or towing or leaving the pavement. They would be better served by commuter vehicles, but all they've ever driven is show trucks and learning new things is scary. If they did attempt to do real work, they would quickly realize the bed can't hold a standard sheet of plywood.

The important thing is that they FEEL like they have capability at their fingertips, even if this is obviously an illusion to people who actually use those capabilities.

[nradov]

It's possible you found a bug in Excel somewhere but I guarantee that when working with large files it's generally faster and more reliable than the competition. I have successfully worked with files way larger than 20 MB which make competing products such as Apple Numbers or Google Sheets completely choke.

[yummypaint]

Compare with gnumeric, that's the only spreadsheet I can recommend. I have files with hundreds of plots, non-trivial fits, and logic. 100% of everything works always and it's much faster.

I have a direct comparison because the exact same computations were previously implemented in excel and libre office. Both dropped plots from files, had straight-up reference bugs once things got large, and would regularly crash attempting to render dozens of tiled plots in 4k. The idea of using Google sheets for this is laughable.

[MegaDeKay]

In my past life, we had a mix of Linux users using LibreOffice and MS Windows using Office. It was indeed at times painful, especially when LO content had to be merged into a Word doc.

But too often I think people just think of Word vs Writer but we're talking the Office experience here. Calc is a poor man's version of Excel: I've found it slow with many rows of data and crash-prone (Office is surprisingly solid). Then there is Visio vs Draw. Use Draw for anything complex and you're going to have a really bad time. Us Engineering folk would put together Visio documents all the time and embed them in lengthy technical documents and proposals. Trying to do this in LO is a road to ruin. The Linux folks would either draw diagrams with sticks and boxes or get somebody in Windows to make something decent in Visio.

What we ended up doing was giving a Windows VM with Office on it for those Linux users that needed to produce documents and the like.

[Suppafly]

>What we ended up doing was giving a Windows VM with Office on it for those Linux users that needed to produce documents and the like.

and at that point, you might as well just use windows, you aren't getting any advantage out of linux and are spending a bunch of overhead managing it.

[nradov]

The 80/20 rule doesn't apply here because it turns out the "20" is different for everyone. If you take away one critical piece of user or organization's workflow then it doesn't really matter that everything else still works.

For LibreOffice there is still a huge functionality gap in VBA support. This is mission critical in a lot of places.

[rpdillon]

I'm just saying 80% of spreadsheet users' needs are covered by 20% of excel's features. We can adjust those percentages, but I do think the principal holds that some features are used by a broader population of others, and Calc tends to focus on those features.