About 10 years ago, I inherited a not-officially-licensed document class for Ph.D, plus a sample skeletal thesis. theses in my alma mater's CS department. I've been maintaining and improving it on GitHub for the past several years, but - have neglected to declare a specific license for it; it has no LICENSE file nor a license block in the sources (and didn't have them when it got to me).

My questions:

  • Should I specify a license for it, explicitly, given that it hasn't had one so far?
  • What should that license be? LPPL? GPL? BSD?


  • 3
    I'm assuming that you've already determined successfully that there never was any license-type caveats to the software you've inherited and have been maintaining generously. Do you want to make the license as liberal (i.e., nonrestrictive) as possible? If so, do also consider the MIT license.
    – Mico
    Jan 29, 2022 at 18:50
  • @Mico: 1. It's going to be more like "affirmed convincingly" rather than "determined successfully", I'm afraid. The origins of this class are shrouded in the mists of time by now. 2. Why MIT rather than BSD? And - why any of those two rather than LPPL or GPL?
    – einpoklum
    Jan 29, 2022 at 19:06
  • 1
    I don't think this is really on topic, if you feel happy asserting control over the code use whatever licence you like, if you do not feel you have that authority you can't pick a licence anyway, Jan 29, 2022 at 19:07
  • IANAL. What little I think I know about software licenses (especially the nonrestrictive ones) comes from reading articles on Wikipedia and similar (hopefully reasonably reliable) sources. In this article, one finds claims that (a) the MIT license is the single most commonly used license on GitHub and (b) that MIT could be ever so slightly less restrictive than BSD. For sure, though, I am in no position to evaluate the correctness of these claims on my own.
    – Mico
    Jan 29, 2022 at 19:29
  • Unrelated to your question, you may consider: 1) adding embedded documentation using dtx; 2) including a suite of tests for the class behaviour. Here is a thesis class I maintain that demonstrates both - not trying to self promote, just in case you're interested... Jan 29, 2022 at 19:59

2 Answers 2


You should always give any code you publish a license. If you don’t, it’s illegal for anyone to download, extend or use it. You know you aren’t a copyright troll who’ll search for anyone who used the code who gave them, catch them using it, and send them a demand letter threatening a $150K lawsuit unless they pay you for a retroactive license. But other people don’t.

If you want a very permissive license, which basically says only that people have to give you credit and cannot remove the license, you want either the MIT license or the 3-clause BSD license.

Most of LaTeX is under the LaTeX Project Public License, which limits the ability of anyone but the official maintainer of a package to modify it. This has caused problems, however, when the official maintainer hasn’t been cooperative.

  • +1. One could also mention that the MIT-0 and BSD-0 license variants do not even have an attribution/credit requirement.
    – Mico
    Jan 29, 2022 at 19:35
  • 1. It's not illegal. 2. Licensing can be implicit in publication. 3. We're not talking about the US here.
    – einpoklum
    Jan 29, 2022 at 19:45
  • @Mico I would read, e.g., “The above copyright notice and this permission notice” in the MIT license to include the original copyright attribution.
    – Davislor
    Jan 29, 2022 at 19:46
  • @Davislor - The "basic" or "standard" MIT license does indeed contain the clause "The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software". In contrast, the MIT-0 variant, aka the "MIT No Attribution" variant, does not.
    – Mico
    Jan 29, 2022 at 19:58
  • 1
    @Mico I was thinking of the classic MIT license. Maybe I should add links.
    – Davislor
    Jan 29, 2022 at 20:00

If you plan to distribute the class officially in any way (e.g. uploading to CTAN, or having your university officially endorse it) I think the LPPL is useful because it prevents others from modifying the code and distributing it under the same name, which can lead to version issues.

  • 1
    I don't think this is CTAN material... but point taken regarding the restricting use of the name. On the other hand, the name I use right now is "presumtive" - I'm using the institution's name (iitthesis for "Israel Institute of Technology"), so the university might be able to claim rights to this or a similar name regardless.
    – einpoklum
    Jan 29, 2022 at 21:13

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .