I have ca. 190 test cases on which I run my software and compare the md5 sums of the resulting PDF. If they are not the same, I create a PNG for every page and compare visually with imagemagick.
The trick is to remove all random stuff from the PDF (like ID generation or such).
This takes about 3 seconds on the M1 Pro laptop. I think this is very much okay.
I have a related feeling about TeX. It has superb output quality but the programming is awful. When LuaTeX finally arrived a few years ago, it was possible to do almost everything you have done before in the TeX language (starting with \backslashes) in Lua.
I have (shameless plug) created a database publishing software using this technique (https://github.com/speedata/publisher/). Once in a while I have to use LaTeX and it feels a bit old school to do the macro programming.
My next project is to rewrite the TeX algorithms in Go - see https://github.com/speedata/boxesandglue. Already usable but not TeX like in any way (this is just a library, not a frontend software like TeX)
I feel the same way! I also feel like the LaTeX crap on top of Plain TeX is crap and it fills me with rage when I try to find info about TeX and always get back something assuming LaTeX.
Various senses that concur.
It has, in addition to the line-breaking algorithm, a global page-breaking algorithm ... and for the rest may I suggest reading (it's brief) section 4.2 of GNU TeXmacs, A free, structured, wysiwyg and technical text editor, J. Van der Hoeven, Cahiers GUTenberg, no. 39-40 (2001), pp. 39-50, that you can find at the official website of the journal at http://www.numdam.org/article/CG_2001___39-40_39_0.pdf
Did you try to enforce the AGPL somehow? In the sense that the user that puts it on a (web)server must provide the end user the source code? (The 'A' in AGPL).
Not really. It's an application you install to your internal network and most probably do not make it accessible publicly. I have no way to know who and where has installed it. I assume that the app is forked in Github and the changes to disable license key check are also committed to that fork, so this would make it valid in terms of AGPL.
> It's an application you install to your internal network and most probably do not make it accessible publicly.
If people don't make it accessible publicly, then the AGPL isn't really imposing any obligations on them, even if they modify the code. It's unclear to me exactly when what you did would lead to someone paying when they otherwise wouldn't.
A lot of people do not care about licenses at all. No matter what you list as the license, they will do so if they can download and run it. Over the years, I have tried this multiple times where I provide the same code under a copyleft (free) license and commercial or permissive license (paid). Not a single time I've had anyone come to me and pay me to make it around the copyleft license. Any time someone has purchased the other option, it has been because of potential support options, etc.
So in EmailEngine, I added a license check. The code is always the same, but if you do not provide the license key (you need to pay to get that key), EmailEngine does not work correctly. Specifically, it stops processing email accounts after 15 minutes of runtime. You have to restart the application to get it functioning again. You would have to modify the source code to overcome this, but EmailEngine is distributed as a binary. To edit that code, you have to download it from Github, modify it and figure out how to run it. It's all straightforward, so it's not an obstacle for anyone with enough dedication. For others, it is easier to pay for the license key than to figure out how to circumvent it.
Did they ever detail what happened, they kept saying nothing has been lost and they were ready to return the site, any minute now, I gave up looking after a few months assuming they'd had some sort of deletion accident they weren't prepared to confess.
IMO contributors deserve to know what happened still. I think there was a lot of agreement that a torrent backup should be made available so the responsibility for the OCAL community's work didn't have to rest with one (?) person who seemingly didn't have time to manage things.
I give away my database publishing software (https://www.speedata.de/) for free and make a living with offering services (helping companies producing high quality product catalogs and datasheets etc. using the software) Its not much, but I live a happy live.
I spend about 1/2 of the time doing services and 1/2 hacking on the software. Quite often, customers are paying for new features in the software so it is paid hacking. Great!
I started with a closed source software but I opened it shortly after. So I didn't start with this in my mind, but since I love open source software, I had the feeling that I should open it and try my luck. I love the idea of the GNU projet.
I really like Lua and hacking things in Lua, so this editor looks like its what I need. But there is one show stopper, that I have no idea how to avoid: scrolling horizontally on my Mac is terrible slow, and I have lots of files with long lines..
The author and the communtity around it is great, all questions that I had have been answered quickly. So every few months or so I download the newest package and look if anyting has changed meanwile.
Yes, I've had an exchange with the author about this, and he said, it might be possible to tweak this, but his is buried in some obsucure configuration.
Depending on how long the lines are, wrapping may turn out to be a pain in the neck because (part of) one or two lines take up the entire pane. So, though wrapping long lines works sometimes, it genuinely is impractical at other times. Also, like the other poster says, it's a matter of preference too. If someone likes to edit a line while having an eye on the lines below and above it, wrapping doesn't help.
I've always avoided this by formatting source code to fit in 80 chars, using "long" data format where possible, and just not editing big data files interactively (I use sed or awk).
Not many comments about privacy. This is how I see it: You are identified when you enter the shop and amazon knows exactly the products you buy.
If you have the choice to buy or not to buy at the shop, that's fine, it is your decision. But let's imagine that in the not so far future, all shops in your neighborhood are like this. No way to go shopping whithout given exact trace of you, your location, the stuff you buy, the time you buy, the amount of food etc.
We all know that too much data is not good for us (yes, I am looking at you, my government).
While I like the idea of not standing in a line and wait, I really wish that these shops offer a prepaid anonymous card for those who don't want to be totally tracked.
There are a lot of social issues that a system like this might entail, in addition to privacy:
1. Not everyone has a smartphone.
2. Not everyone has an electronic account with which to pay (or for practical reasons, they can't use it frequently).
And the people that fall into those categories aren't just fringe. Parents give dollar bills to their kids & send them to the store to fetch food. People without established credit might have to use cash for purchases, etc.
Probably - I don't know the procedure. I do know most people consider debit to be a less secure (for the consumer) form of payment than credit or cash though.
I think the privacy implications for a shopper at Amazon Go are the same as a shopper who shops at Tesco with a credit card. The company can use your credit card to track exactly what you buy already and many people seem to be happy with that (even going so far as to sign up for loyalty cards).
But it should be possible to a) use a prepaid credit card or b) pay with good old cash. That way it is still possible to buy without getting tracked (while all the others that don't care use the loyalty card).
I agree it should be possible but unfortunately the technology doesn't allow it. In this case I would give the analogy of online vs offline shopping. Ideally we would be able to buy goods on amazon.com with cash (or perhaps bitcoin) but the technology doesn't make this feasible.
don't know about tesco, but I know about Costco. You need to provide your membership card every time you buy. So Costco definitely knows your complete shopping history.
Also any online shop (Amazon included) has a complete history of everything you've ever bought.
For brick&mortar shops you would be a fool to assume they don't mine your shopping history based on your credit card (or some sort of hash that's computed from your CC#).
I have ca. 190 test cases on which I run my software and compare the md5 sums of the resulting PDF. If they are not the same, I create a PNG for every page and compare visually with imagemagick.
The trick is to remove all random stuff from the PDF (like ID generation or such).
This takes about 3 seconds on the M1 Pro laptop. I think this is very much okay.
Links: https://github.com/speedata/publisher/tree/develop/qa (the tests) https://github.com/speedata/publisher/blob/develop/src/go/sp... (the Go source code for the comparison)