（评论）

（评论）
(comments)

原始链接: https://news.ycombinator.com/item?id=40284219

本文讨论了名为 pyspread 的特定电子表格工具的用法和首选项，重点关注其与 macOS 的兼容性以及与 Microsoft Excel 相比的易用性。作者认为，多功能性和迎合各种操作系统可能会导致开源项目维护人员的倦怠。他们建议大多数专业开发人员可能会使用 macOS，因为它与 BSD 系统相似，这使得 pyspread 成为一个有利的选择。不过，他们也承认，美国初创公司可能会提供灵活的硬件选择，从而导致 MacBook 在开发者中广受欢迎。文中提到了 Excel 的问题，例如使用字母顺序的列名称对于较大的矩阵来说变得令人困惑、缺乏依赖性跟踪以及确定列偏移量的麻烦。另一方面，Pyspread 提供了更简单的数字标签，尽管它缺乏真正的电子表格功能，例如依赖性跟踪。作者对旨在通过“增值”将组织困在各自云服务中的专有软件表示担忧，并批评 Excel 和 Microsoft Office 根据 GNU 通用公共许可证版本 3 获得许可，从而阻止其包含在 FreeCAD 等其他开源工具中。他们强调开源贡献对职业发展的重要性，并通过开源平台分享个人参与和成功的经验。此外，他们建议针对所需领域内的特定组件，以帮助获得专业知识并提高就业能力。

I'd like to give praise for the "Target User Group" section on the homepage.

Not only does it say what users the app is for, but also who it is NOT for. I think this kind of information is invaluable in deciding whether or not to use or suggest an app.

I can understand if app developers want EVERYBODY to user their app (whether or not its the best for the job) or if the app developer just doesn't want to take the time to write out who the app is NOT for. But I will praise those who do include that information.

After reading the parent comment - I thought to myself “so what”.

But after reading it on the product page - I fully agree: seeing clearly the targeted personas and out of scope usage significantly elevates my trust in the product and the team behind it.

Have not used the software, but now I want to try

I like it and I'll take it a step further. I think this is important information for a developer to admit to THEMSELVES at least. Being all things to all people is a route straight to burnout for an open source project.

pip install pyspread works just fine on macOS (assuming you have pip and python installed, which anyone who would be interested in having a python-enabled spreadsheet would have).

Not quite true, as others have pointed out. However, the installation instructions say this:

> While there have been reports that pyspread can be used on OS X, OS X is currently unsupported (can you help?).

I think that would make many non-technical users uncomfortable.

I wish Stackoverflow included "daily operating system" in their surveys, but I'd wager that 50%+ of professional developers use MacOS. For many reasons probably, but for me it's simply because it's a nice BSD system that translates well to Linux servers (prod).

    > but I'd wager that 50%+ of professional developers use MacOS

Ok, I will take that bet against you. If you think it true, then you are living in a bubble. Probably 90% use MS Windows with no choice, because they work in a big enterprise that uses MS Office and Exchange/Outlook. For enterprise email, is there any competition at this point? I don't know any.

In that case, I would take your side of the bet! US startup devs appears (from HN) to have pretty liberal hardware policies. If choosing a laptop, many will choose the best hardware, which is currently a MacBook.

I found a typo in the "Target User Group" section.

Original Sentence:

> Furthermore, Peter is worried about algorithmic errors and calculation inaccuracies is his complicated tables.

Correction:

> Furthermore, Peter is worried about algorithmic errors and calculation inaccuracies in his complicated tables.

I would really like to see a distribution which puts all the best alternative software together:

- pyspread for a spreadsheet

- LyX for a word-processor

- OpenSCAD for a 3D modeler

- TkzEdt (or ipe) for 2D drawing

&c.

(and I'd be interested in suggestions for similar software for other tasks, esp. presentations and database work)

KiCad for electronics development.

Blender for 3D graphics.

Gimp for photo editing.

Inkscape for illustration.

Though I'd recommend FreeCAD over OpenSCAD as a 3D modeling tool for most users as an alternative to traditional CAD tools...and therein lies the distro problem.

Gimp is really not great for photo editing IMO - it really shines at photo manipulation. (i.e. it is Photoshop, not Lightroom.)

For RAW development (a la Lightroom), DarkTable and RawTherapee exist. I've only tried the former, and found it incredibly difficult to use.

I also prefer Krita for digital illustration (though maybe that's unfair - it's best at digital painting specifically.) Of course, if you need vector support, Inkscape is the obvious answer.

But why do you even need a distro to begin with just to ship certain software set? Install Arch/Gentoo and install whatever is your preferred software of choice, or hell, you can even do that on any other distro.

The value a dedicated distro provides here is that you don’t have to do the legwork to research and find the best tools.

As power users we may want to do that ourselves, but a lot of people place value in having that curation done for them.

Isn’t that basically package groups (or whatever the district-specific terms are)?

Eg Im pretty sure Ubuntu Desktop has some kind of “Productivity” package group that includes a word processor and spreadsheets and an email client and what not. I’m pretty sure it’s selected by default when you do a full desktop install. I don’t recall what the actual software is, but I would imagine LibreOffice.

I would agree with OP that it doesn’t really make sense for a distro, though. People really want to “make a distro” for some reason so we end up with silly shit like Kubuntu (Ubuntu… with KDE pre-installed).

My general rule of thumb is if I can point the distro’s OS package manager to the distro’s upstream (ie Ubuntu for Kubuntu, or Debian for Ubuntu) and everything works or mostly works, it should be a script or apt repo and not a distro.

There are way too many “Ubuntu but with a different default DE” distros that could really just be a modified install ISO or post-install script.

OpenSCAD is really good for parametric designs. One of the first things I designed in OpenSCAD was a bicycle sprocket where you could input how many teeth you wanted, link length, how thick, etc. and OpenSCAD would generate the sprocket according to those parameters. I recall seeing a while back that blender was adding support for this kind of thing so I'm sure it's possible, but it was very intuitive in OpenSCAD.

For any large project I'm sure Blender (or FreeCAD) would be a better choice, but as someone with some programming background just starting out, OpenSCAD feels way more accessible.

I've tried a lot of 3D software over the years, and OpenSCAD has been the one I've been most successful with (along w/ Carbide Create, but I work for that company....)

Currently working on a library which makes the two work together:

https://github.com/WillAdams/gcodepreview

(which is currently quite primitive/basic, and even when fully developed is not likely to be used by anyone else)

Blender wasn't originally designed to be used for CAD type applications. It's possible of course, but you're likely better off with something specifically meant for precision CAD models. Specifically for 3d printing, I don't think blender is ideal.

I think most programmers like OpenSCAD - what better way to make parametric CAD models than by writing code?

A lot of cad software actually has an editable history: you can go back 100 steps and modify a dimension or a step and have the rest of the model updated to take that new dimension into account.

I wouldn’t work with any cad software that didn’t have that ability.

This is called parametric design. Since CAD programs do not operate on meshes, this allows you to e.g. change the dimension on some part and have the rest adjust accordingly.

How well this works depends on the situation, if changing a parameter causes new faces to appear on the object this is usauayvwry difficult to handle, even for commercial CAD programs.

I see it uses numerical designations for both the columns and rows allowing indexing like a 2D matrix. Nice. One of the many annoyances of Excel is the alphabetic columns that make even less sense beyond 26.

Baby mode?! As a full grown adult, I find the cognitive load of the base 26 arithmetic of "what column is an offset of 8 to the right of AW" to be exactly what I needed to fully forget about what I came there to do in the first place

Yes. But many parts of excel don't support those. E.g. conditional formatting just doesn't deal with names.

Also, when you need names, it is a sign to reach for a proper programming enviroment like Delphi or Lazarus.

some of us have some logical dyslexia, so when I see a tuple of numerics, even if you say (this is the row, this is the column, this is the table), when I'm manipulating references, or finding a tuple in the dark, I easily mix up the order of the bits, because lets face it, it's arbitrary.

So, while your logic makes sense, it voids humanity in a rather stark way which I suppose is fine, but it's not like I'm incapable of programming. I just have to be careful about how I name variables and use keywords in python more often than args.

It seems to not be a true spreadsheet with dependency tracking and instead it recomputes everything after each change. Whis probably kind-of obvious as how to track the dependencies for arbitrary Python code is not that obvious. But without resorting to some static analysis magic one could just track accesses to the magic S[] during the excecution of the cell, its value cannot depend on anything else than what was accessed during the evaluation.

One issue with this is that arbitrary Python code can have arbitrary side-effects.

Your suggestion reminds me a lot of fine-grained reactivity like in SolidJS, which makes sense, since spreadsheets basically operate on reactive programming. Some great articles by Ryan Carnatio on the topic.

The side-effects thing comes in if a user puts in some side-effect in a dependent cell, which is equivalent to adding side-effects in a memo in reactive-speak.

That gets broken by constructs like sum(all of column 5) when new rows or columns are added.

(It's very similar to the problem of locking in databases)

I have been using grist [0] for a number of months. Self-hosted. Uses Python and has defined all common Excel functions as Python functions. Has a web-api so it can be used as a kind of database. Very nice for my small use-cases where Excel would have fit the bill. I have not stress tested it and I know there are row limitations without going to the enterprise edition.

0: https://www.getgrist.com/

Not actually "in Excel", though. The Python code runs on Microsoft's servers (they say in the introduction) and Excel is just a client.

There's no reason they couldn't embed CPython in Excel, but maybe the intention was for the online version of Excel to have feature parity without having to compile Python to JavaScript?

the intention is to lock in orgs to their cloud services. This is a value-add. They really know that Excel, Word are "feature complete" and the only way they're going to make money on it is by harvesting and locking in the users.

Shameless plug: If you have bigger data sets, check out https://rowzero.io

We scale up to hundreds of millions of rows and have native Python support.

You can define functions in Python and call them as formulas from any spreadsheet cell. We seamlessly marshal Pandas dataframes from Python land to spreadsheet land and back. [1]

We're also hosted and support real time collaboration like Google Sheets. We reimplemented the Excel formula language. We connect directly to Postgres, S3, Snowflake, Redshift, and Databricks. And the first workbook is free.

[1] https://rowzero.io/docs/code-window

Shameless suggestion: Instead of one free workbook, make unlimited free workbooks, but unusable for corporate settings. I'd recommend binary sharing:

- Only I have access (private / personal use)

- The universe has access (open-source use)

But one free workbook still beats the typical 30-day trial.

Shame it's GPL3. That counts it out of being included in FreeCAD, which I can't help but feel would be an improvement on the current spreadsheet workbench.

I'm curious how it compares to emacs' calc / spreadsheet mode in org.

Looks like it is

a) a stand-alone X11 app (not easily runnable on macos) b) using python instead of elisp.

Is that roughly accurate ? I can't easily understand from the post.

This looks a very good idea.

This would be an ideal place for a semi-technical user to contribute documentation. Good starter open-source contribution. Without that, it's hard to figure out if it would work for me, let alone give feedback.

Some of the motivation for this comes from how often devs want to contribute to open source but are intimidated by how difficult the barrier of entry is, particularly for large projects. It's surprisingly hard to find a good list of projects that a beginner or even intermediate programmer can substantially contribute to. The ones that do exist tend to have the low hanging fruit plucked pretty quickly.

A few points:

- Most of my career was made by being the author of one popular open source platform which happened to do well.

- I've recruited people based on open-source contributions. If I want an expert in [X], finding someone who contributed to [X] is a good bet.

- The flip side is I've made (minor, helpful) contributions to many projects in part for exposure. My name is in the commit list of many systems in domains where I have wanted to work.

- Many mid-sized contributions look good on a resume, especially for a junior developer. Indeed, I've made one case to promote someone based, in part, on contributing to a library we were using (even if only tangentially).

If you want a job in e.g. network security, find something in a firewall, anonymzing proxy, packet sniffer, or whatnot, and make a PR. It's often quick, easy, and helpful. A corollary is you do actually learn a lot about a system by contributing.

I have no axe to grind here, but I think the cynicism is unwarranted.

I love open source. My cynicism isn't about open-source, but about the OP's first post being "these docs suck, snaps fingers maybe one of you non-devs can work on it."

Good life lesson: You don't know until you ask.

That's more a salesperson mantra than SWE, where for every 10-100 people you ask, someone buys something. However, I've raised money many times simply by shamelessly asking.

Second good life lesson: Don't assume things about others.

People who make $1/day, $10/day, $100/day, $1000/day, $10,000/day, etc. have fundamentally different priorities and motivations. Ditto on many other axes. Good synergies are leveraged working across such differences.

So what, who want raw wild naked large exposure? If you don’t get an army of free PR specialists, lawyers and body guards to protect you and your beloved ones h24 every single day for the rest your lives, it’s an obviously net negative situation.

>Every dev here loves doing volunteer work as a favor.

Devs are expected to contribute to FOSS, write free educational blog posts about technology, and fix your friends computer!

Ask a doctor why something hurts and they tell you to come into their office and insurance will bill you.

This sounds made up tbh. I mean, I agree it's common to feel this "pressure" but the day you discover the pressure isn't real is the day you level up. And I charge plenty for fixing peoples' software pains.

I do open-source and write blog posts to satisfy my own desire to publicize my work, and for exposure, and to put my skills on display. Not as a race to the bottom. I could have had a career without those things.

Python is a bad language for this, because one-liners are awkward. Multiline code is possible but ugly due to indentation-based syntax. Brace-based languages would be far more suitable here.

I tried to install, and discovered yet again why I fell out of love with Python.

Executing:

pip3 install -r requirements.txt

gave this error:

" Traceback (most recent call last): File "/tmp/tmprhazpypn", line 126, in prepare_metadata_for_build_wheel hook = backend.prepare_metadata_for_build_wheel AttributeError: module 'sipbuild.api' has no attribute 'prepare_metadata_for_build_wheel'

...

ERROR: Command errored out with exit status 1: /usr/bin/python3 /tmp/tmprhazpypn prepare_metadata_for_build_wheel /tmp/tmpl72s0sfd Check the logs for full command output."

I'm on Linux Mint 19.

Use pipx to install applications instead. It even works on Windows.

  > pipx install pyspread
      installed package pyspread 2.2.3, installed using Python 3.12.3
      These apps are now globally available
        - pyspread.exe
    done!

I tried to install it on ubuntu 22.04 and got a different error using pip. I'm not experienced enough with Python to quickly figure out how to proceed. It's a shame it isn't easier to install.

Debian has it out of the box:

    $ LANG=C apt-cache policy pyspread
    pyspread:
      Installed: 2.2.3-1
      Candidate: 2.2.3-1
      Version table:
     *** 2.2.3-1 500
            500 http://ftp.de.debian.org/debian trixie/main amd64 Packages
            500 http://ftp.de.debian.org/debian trixie/main i386 Packages
            100 /var/lib/dpkg/status

Compare this with the comments above re nix where it just ran - although to get a python app to build on nix is a pain but at least only one person has to do it.

You're not really supposed to pip install applications. pip is a developer tool. There is pipx which is great, but I still think it's a developer tool. It's annoying when projects say to pip install stuff. It's never the right thing to do.

Under Fedora 39 I was able to pip install and run it withhout problems, inside a venv environment:

    (pyspread) [xxxx@fedora ~]$ pip install pyspread
    Collecting pyspread
      Obtaining dependency information for pyspread from https://files.pythonhosted.org/packages/a5/e2/19ddb20b46ae46f11102f9095bf5bf00cb28cc79b35f38257f84a98ecd1d/pyspread-2.2.3-py3-none-any.whl.metadata
      Downloading pyspread-2.2.3-py3-none-any.whl.metadata (4.8 kB)
    Collecting PyQt5>=5.10 (from pyspread)
      Obtaining dependency information for PyQt5>=5.10 from https://files.pythonhosted.org/packages/2f/e6/a1f9853e4933c312c6de9c79d126c7d92ef69ae0e53895fb1ceb0ecc77a6/PyQt5-5.15.10-cp37-abi3-manylinux_2_17_x86_64.whl.metadata
      Downloading PyQt5-5.15.10-cp37-abi3-manylinux_2_17_x86_64.whl.metadata (2.1 kB)
    Collecting markdown2>=2.3 (from pyspread)
      Obtaining dependency information for markdown2>=2.3 from https://files.pythonhosted.org/packages/5a/09/a9ef8d5fe4b08bfd0dd133084deefcffc4b2a37a9ca35a22b48622d59262/markdown2-2.4.13-py2.py3-none-any.whl.metadata
      Downloading markdown2-2.4.13-py2.py3-none-any.whl.metadata (2.0 kB)
    Collecting numpy>=1.1 (from pyspread)
      Obtaining dependency information for numpy>=1.1 from https://files.pythonhosted.org/packages/0f/50/de23fde84e45f5c4fda2488c759b69990fd4512387a8632860f3ac9cd225/numpy-1.26.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata
      Downloading numpy-1.26.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (61 kB)
         ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 61.0/61.0 kB 311.8 kB/s eta 0:00:00
    Collecting setuptools>=40.0 (from pyspread)
      Obtaining dependency information for setuptools>=40.0 from https://files.pythonhosted.org/packages/f7/29/13965af254e3373bceae8fb9a0e6ea0d0e571171b80d6646932131d6439b/setuptools-69.5.1-py3-none-any.whl.metadata
      Downloading setuptools-69.5.1-py3-none-any.whl.metadata (6.2 kB)
    Collecting PyQt5-sip<13,>=12.13 (from PyQt5>=5.10->pyspread)
      Obtaining dependency information for PyQt5-sip<13,>=12.13 from https://files.pythonhosted.org/packages/3c/ab/f8f1e970768fcb4ab118d4aabbfcb9b7f781088b71e1f26d813fd51c4701/PyQt5_sip-12.13.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.whl.metadata
      Downloading PyQt5_sip-12.13.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.whl.metadata (504 bytes)
    Collecting PyQt5-Qt5>=5.15.2 (from PyQt5>=5.10->pyspread)
      Obtaining dependency information for PyQt5-Qt5>=5.15.2 from https://files.pythonhosted.org/packages/83/d4/241a6a518d0bcf0a9fcdcbad5edfed18d43e884317eab8d5230a2b27e206/PyQt5_Qt5-5.15.2-py3-none-manylinux2014_x86_64.whl.metadata
      Downloading PyQt5_Qt5-5.15.2-py3-none-manylinux2014_x86_64.whl.metadata (535 bytes)
    Downloading pyspread-2.2.3-py3-none-any.whl (1.7 MB)
       ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.7/1.7 MB 2.7 MB/s eta 0:00:00
    Downloading markdown2-2.4.13-py2.py3-none-any.whl (41 kB)
       ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 41.3/41.3 kB 169.4 kB/s eta 0:00:00
    Downloading numpy-1.26.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (18.0 MB)
       ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 18.0/18.0 MB 4.1 MB/s eta 0:00:00
    Downloading PyQt5-5.15.10-cp37-abi3-manylinux_2_17_x86_64.whl (8.2 MB)
       ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 8.2/8.2 MB 4.8 MB/s eta 0:00:00
    Downloading setuptools-69.5.1-py3-none-any.whl (894 kB)
       ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 894.6/894.6 kB 3.1 MB/s eta 0:00:00
    Downloading PyQt5_Qt5-5.15.2-py3-none-manylinux2014_x86_64.whl (59.9 MB)
       ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 59.9/59.9 MB 1.8 MB/s eta 0:00:00
    Downloading PyQt5_sip-12.13.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.whl (360 kB)
       ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 361.0/361.0 kB 2.0 MB/s eta 0:00:00
    Installing collected packages: PyQt5-Qt5, setuptools, PyQt5-sip, numpy, markdown2, PyQt5, pyspread
    Successfully installed PyQt5-5.15.10 PyQt5-Qt5-5.15.2 PyQt5-sip-12.13.0 markdown2-2.4.13 numpy-1.26.4 pyspread-2.2.3 setuptools-69.5.1

    [notice] A new release of pip is available: 23.2.1 -> 24.0
    [notice] To update, run: pip install --upgrade pip
    (pyspread) [xxxx@fedora ~]$ python3 -m pyspread

Stop using debian-family.

Dealing with outdated/bugs is just a regular day under the guise of 'stable'.

The only reason you even are using Mint is because conical did a marketing trick sending free CDs in 2000s.

stop trying to use the system python for application development. You will break your OS and you will make your code machine specific. Go read up on Virtual Envs , you were the problem here.

Python has like a dozen competing solutions in their package management space. Python is clearly the problem, not the user.

Their docs don’t say that, they advertise exactly what the GP ran. Python needs to unfuck its ecosystem.

Yeah they kinda screwed the pooch on that.

That being said, venv is the official solution, and it works fairly well. pipenv/pipx are some alternatives...but yeah.

What exactly is your system installation for if you aren't supposed to install anything to it? The package system should be able to attempt installing via venv or be able to block uncompatible packages with sane error messages.

If venvs were a reasonable solution here, presumably the pyspread authors would have included them in the installation instructions. Instead they just say to "pip install", which predictably does not work for most people. It sounds a bit unfair to tell someone that "they're the problem" when they followed the official installation instructions to the letter. I would agree in classifying this as another case of python developers being unable to provide a working way to install their software (if code that the average person can't even get to execute can even be called "software".)

venv not always a panacea for system python

  $ docker run --rm -it python bash
  [...]

  root@211646f0fa99:/# which python
  /usr/local/bin/python

  root@211646f0fa99:/# python -m venv .venv

  root@211646f0fa99:/# source .venv/bin/activate

  (.venv) root@211646f0fa99:/# which python
  /.venv/bin/python

  (.venv) root@211646f0fa99:/# ls -l "$(which python)"
  lrwxrwxrwx 1 root root 21 May  7 14:47 /.venv/bin/python -> /usr/local/bin/python

Correct, virtualenv isolates package installation but doesn't handle installing or managing arbitrary Python versions.

Node.js is similar with npm.

Is there something like nvm but for Python?

> stop trying to use the system python for application development.

Umm... who exactly makes sure the app runs correctly with the version of python installed by the system?

I, for one, aren't going to bump up a python version to run a single app in a virtual environment because some dev decided to use 'python nightly' to code against.

I think one major difference between spreadsheets and Jupyter notebooks is state/dependency graph. In a notebook, cells don't depend on each other. They can be run many times, and the order they're run in matters. This is one of the major gotchas with notebooks. With spreadsheets, they're generally deterministic. There is no notion of "cell order", cells are recomputed automatically depending on the dependency graph between the cells. This results in a pretty big user experience difference.

Yes in that it doesn't appear to use a dependency graph to track recomputes; it recomputes everything on a cell change (according to another commenter anyways). But the UX effect is still the same in that the user never has to think about cell order, and every cell is always consistent with each other.

the magic power of spreadsheets is that they encourage improvisation, and it probably applies to that one.

you have only one data structure (the 2D table), data types are super-weak, there are no variable names... all of this guarantee a maintenance nightmare, and rightfully scares developers. But it's also a very low barrier to entry. You've got data, you paste them into the grid, and you start toying with them, before having figured anything about them.

That's an amazing superpower, when targeting non-developers, and that's why Excel is the most used programming language over the world, by far: it's probably got an order of magnitude more users than there are trained developers in the world.

and _that_ is why I'm still very sad that Lotus Improv didn't make it in the marketplace --- gathering all the formulae into one pane was _incredible_ for organization and providing a single top-level view of what a spreadsheet was doing.

I really wish Flexisheet would get to a usable point, or that Quantrix wasn't so expensive.

This is of course the correct answer to the question. But additionally, I made the joke because I somewhat Identified with the 'Jack' use case mentioned in one of the pages on the PySpreads site

（评论） (comments)

（评论）
(comments)