On 20 May 2015 at 01:39, Anil Madhavapeddy <[email protected]> wrote: > On 20 May 2015, at 00:07, Daniel Bünzli <[email protected]> wrote: >> >> Le mercredi, 20 mai 2015 à 00:45, Magnus Therning a écrit : >>> I can't seem to get `camlp4` to run to completion when building on >>> ARM. I cloned mirage-decks, added a small deck, and then removed all >>> the others in an attempt to limit the compile time resource needs... >>> but still: >> >> In the past I had similar problems on rpi with camlp4 because it was using >> too much memory and had to manually increase the swap size (see [1]). Is it >> maybe a similar problem ? > > Increasing swap is a definite necessity. I also notice that Magnus > is compiling mirage-decks, which has grown very large recently, and so > has outgrown `crunch`. I usually create a new `slides/` folder with > just my current talk and the `slides/templates/` folder and then compile > that on ARM instead. > > Do you want to try that and see if it improves things on your Cubie, Magnus?
I did miss including a bit of information about the setup: - I've created a 2G swap partition on the system - Before compiling `mirage-decks` I've removed all decks except my own (I made a clone and this is how I remove them: <http://is.gd/ZJ7bbH->) - As you can see I get the same kind of issue with another project (a toy project of mine): <https://gitlab.com/magus/mirage-kvsrv> It somehow feels a little bit silly that the system can handle `opam switch 4.02.1` but fails on something like my toy example above. :P /M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: [email protected] jabber: [email protected] twitter: magthe http://therning.org/magnus _______________________________________________ MirageOS-devel mailing list [email protected] http://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel
