The killer features of LaTeX that does not let me go with typst (Although I like my typst generated resume) as an academic are.
1. Beamer, I create multiple slide decks per week and the out of the box setup that beamer provides with different styles and fonts for different needs are unmatched. The efforts to generate some of this on typst is not there yet.
2. Generating figures using tikz and be able to modify it on the source file. Because I don't bear using GUI tools. And now life is easier that LLM can help you with complex tikz generation.
3. Not that it is actually a point but I am used now to overleaf and I have professional account as CERN member. It is also better on collaboration level and features than typst cloud.
I hope that one day typst will grow into this direction so that I can stop using LaTeX. Until then I have couple of overleaf templates generated for my use.
Maybe you have already checked these, but in case you haven't:
- https://touying-typ.github.io/ Creating slides in Typst
- https://cetz-package.github.io/ CeTZ is a package that allows for drawing with Typst with an API inspired by TikZ and Processing. It also provides plotting and chart libraries and is used in several other packages to create circuit, fretboard and more diagrams
- maybe try getting your team to use version control? You may this that it's a lost cause, but existing version control schemes (like git) work very great for textual formats, including with LaTeX or Typst
These are great suggestions, but did I understand that suggest git as an alternative to overleaf? That's... not at all reasonable, as well as completely missing what problems overleaf solves for which people. Overleaf ist git und the hood. But what it really provides is Google Docs style collaboration on Latex documents.
Reviewing the changes I review in Overleaf in GitHub pull requests would be incredibly painful and introduce a completely new, convoluted and unintuitive (until you're used to it) workflow for collaborative editing.
In the Typst web app, you can link a project to a Git repository. This way you can have real-time collaboration in the web app, and versioning through Git. It's not the same as Overleaf's history GUI, but it's functional.
If you make a lot of slides with latex, then it is definitely worth it to try typst. I have a lot of presentations in latex for lectures and such things, with many animated tikz figures. But the compilation times are huge. At some point it is very time consuming to iterate. With typst, it compiles so fast that you don't have to fear to start a compilation. I finish my presentations much faster now.
Cetz has been working very good for me. I was really unsure that it could replace tikz for my applications. But apparently, as long as you have good geometrical primitives (lines, rectangle, circles, etc) you can do a lot. Also it is much nicer to program and make real functions with typst. It is true, the typst options to replace beamer are still not quite there in comparison, but they are definitely in a very useful state. See for example typst-presentate [1].
One thing I'm missing when making slides with typst is the ability to show short videos or animated gifs. Although to be fair this isn't easy in beamer either.
Typst can actually include gifs, but they don't move for me. I have some hopes that perhaps one could make slides straight in html which could alleviate the issue.
1) I have been using typst to create slides with some success. Adding special features tends to be simpler than in beamer.
2) cetz (https://github.com/cetz-package/cetz) works quite well and is comparable to tikz in complexity and capability. of course, there is more support for tikz, but it is bound to improve over time.
To add on what's been said already on slide decks, another great slide creation package in Typst is touying[1]. I've used it to create my own academic theme[2] for courses or conference presentations.
It was easy for me to roll my own slideshow tool in Typst. I would have never attempted this in LaTeX.
Can't respond to 2 or 3. Used tikz once. I wish there were a de facto programmatic drawing tool (I usually resort to graphviz or matplotlib), but there are a bunch of 80% solutions.
No they are not already there. As others replied, there are somethings similar but not the same feature completeness. I was specific to talk about beamer not saying "slides" in general.
> I acknowledge help from ChatGPT 5 for literature research
as well as checking this manuscript. I swear I actually wrote
it myself
Swearing in a research paper is new thing to me. Might pass arxiv pre-published stage but definitely not a peer-reviewed one. But I doubt this will ever get published.
I’ve seen things acknowledge cricket, people who bring them tea and I’ve seen people coauthor with a dead friend who appeared in a dream. Saying “I swear I wrote this myself” barely registers.
This isn't a published paper nor is it associated with a journal. Given the atypically casual writing style I'm not sure it was ever intended for publishing in an academic journal or to be on par with her other papers.
I am already using fmail3 [1] and before that fmail2 which is also a web wrapper but feels more native to mac than Electron apps. And I think it is written in swift. So I don't know why fastmail cannot do something similar after all these years.
Efficiency has improved somewhat, but also, the cost to manufacture PV is enormously lower than it was then, which means that it's using much less materials and energy to produce. Therefore the energy payback time should be much shorter.
How much shorter depends on the fraction of the energy cost in the first place then. Price is not even an upper bound for energy which is cheap in China, and also production is probably subsidised.
Intensive investigations of Chinese "subsidies" from the US Department of Commerce have come up with embarrassingly little.
The current €0.100/Wp price level (12¢/Wp) combined with the US$0.08/kWh given above does give us an upper bound of 1500 hours of full sun for the energy payback time. At a Californian 29% capacity factor, that's 7 months of operation.
But in fact not 100% of the cost of the solar module is the cost of energy; there are some labor and raw materials costs as well, as well as capital costs and, astoundingly, some profit. So 3 months is probably a better estimate.
Process energy will still be a lot cheaper than .08 US$/kWh in China. I doubt they process glass with electric ovens. OTOH that is primary energy then, which kind of supports your argument. Still other emissions in the silicon supply chain that cause global warming are also relevant. If I find the time, I‘ll look for recent LCA.
I'm delighted to see Fraunhofer's report! I'll read it carefully.
You are probably right about the glass. Processing glass in electric furnaces has been a well-established technology for many generations, but it is more expensive than fossil-fuel-fired furnaces, and is used for making more expensive kinds of glass. But the glass in solar panels is an especially cheap kind of glass.
Coal is probably the cheapest form of energy.
https://www.coal-price.com/ says that currently in China coal is on the order of US$100 per tonne. Coal has up to 33MJ/kg, so that works out to 0.3¢ per megajoule, or 1.1¢ per kilowatt hour, or more if the US$100/tonne coal isn't the best anthracite. You probably can't literally run a glassmaking furnace with a coal firebox, because of the contaminants from the coal, but you almost certainly can't run it on cheaper energy than that. If 100% of the cost of the solar panels were coal energy instead of electric, that would extend my upper bound out from 7 months to a bit over 4 years.
This seems to be in rough agreement with Fraunhofer's more informed estimate that you cite.
The only Electron app that is showing the green light is Siyuan which they updated today. Everything else if affected. But most important I deleted some apps I don't know why I was keeping.
> That said, I love how open the astronomy community is with their code and data. I wish other fields would follow their lead, but given the incentive structure, they probably won't.
CERN also provide a lot of open physics data from various experiments and is keep adding large amount each year [1]. Of course this still a needle in the haystack but still more than any individual researcher can ever process.
Even CERN would advice everyone to use ad-blocker [1] for a safer internet experience. I am sure ads as it is today wasn't part of the web plan when it started.
Guess nowadays they recommend everyone to use Firefox or some other non-crippling browser then also?
I helped my wife with something the other day, noticed the ads everywhere, while I was sure I had installed uBlock for her in the past. Went to the Chrome's addons page, and Google apparently is automatically disabling uBlock and calling it unsupported, yet you can enable it until next time you restart Chrome. But seems Chrome is actively trying to get rid of adblockers lately.
I don't, for that I usually rely on search engine or autocorrect. I hate the fact that it is not pronounced as it is written. Because I always go with "Wenesday" on my own.
I don't think it is true that GPT-5 is much better than Claude 4.1. And more important is that light year is distance measure and I am sure openAI data centers is still on earth.
1. Beamer, I create multiple slide decks per week and the out of the box setup that beamer provides with different styles and fonts for different needs are unmatched. The efforts to generate some of this on typst is not there yet.
2. Generating figures using tikz and be able to modify it on the source file. Because I don't bear using GUI tools. And now life is easier that LLM can help you with complex tikz generation.
3. Not that it is actually a point but I am used now to overleaf and I have professional account as CERN member. It is also better on collaboration level and features than typst cloud.
I hope that one day typst will grow into this direction so that I can stop using LaTeX. Until then I have couple of overleaf templates generated for my use.
reply