|
From: Raymond T. <toy...@gm...> - 2022-07-19 00:55:04
|
On Mon, Jul 18, 2022 at 4:51 PM Leo Butler <Leo...@um...> wrote: > On Mon, Jul 18 2022, Raymond Toy <toy...@gm...> wrote: > > > Yes, that's one problem. > > > > I'd really like to change the behavior so that only the user's home > > directory is searched. I don't know of any other program that looks for > > its init files in the current directory. They're always in some fixed > > location in the user's home directory. > > FWIW, I use the current behaviour. On the other hand, one can set the > --init command-line flag, so changing the default should be fine. > Yeah, that's why it's not a huge problem. But thanks for letting me know that there are people who want to pick up the init from the current dir. I don't usually do that. But I also don't want to break things, since it's been this way since forever, I think. > > Note that building maxima as a regular user can run into bad behaviour > during `make check' regardless. In this case, maxima will load the > user's init file--which is almost surely a mistake. The only > short-circuit that I have discovered is to set MAXIMA_USERDIR=/dev/null. > make check should probably run maxima-local with the new --no-init flag. I'll look into it. We should probably make sure that when building locally, we don't load any user init files. It can cause all kinds of strange things to happen that would be hard to figure out. > > Leo > > > > > On Mon, Jul 18, 2022 at 10:55 AM Gunter Königsmann <gu...@pe...> > > wrote: > > > >> I would even deem that as a security-relevant bug: Trick someone into > >> downloading a malicious ini file and then make the same user open a > >> harmless .mac file hoping it will be downloaded into the same folder. > >> Perhaps there are easier scenarios for exploiting that feature, though. > >> -- Ray |