Re: one idea
"Ethan Grammatikidis" <[email protected]>
| Newsgroups | gmane.linux.lib.dietlibc |
|---|---|
| Message-ID | <[email protected]> |
On Thu, 8 Jan 2009 15:19:23 +0100, "Felix von Leitner" <[email protected]> said: > > I tried to build last openssl with > > make CC='diet -Os gcc -nostdinc -s' > > and it fails. The problem was that Makefile expected one > > variable for CC. > > The openssl build system gets more broken with each version. > It's a tragedy, really. > > I just sent them a bug report about this. They also have .../lib/ > hardcoded which is wrong on amd64 systems. > > What a crock. > > I don't think we should try to work around their brokenness. Their > build system is faulty, it's their responsibility to fix it. I emailed > them a fix for their Makefiles. > > Felix I've had problems building openssl in a quite different system too. I've lost the email, but didn't someone mention MatrixSSH & MatrixSSL? I looked it up, it claims 50kB binaries but I'm pretty sure it's not GPL, there is talk of evaluation licenses. If only SSH is needed, how about Dropbear? It is somewhat larger than MatrixSSH; the Dropbear website quotes 110kB statically linked with uClibc. Perhaps it would be smaller with dietlibc. It does however fork and require an external shell, unlike MatrixSSH Dropbear: http://matt.ucc.asn.au/dropbear/dropbear.html As to the library, I don't know if porting applications to GNU TLS is any useful suggestion? Alternatively there is yaSSL, which I didn't know about until just now. GNU TLS comparison with OpenSSL: http://www.gnu.org/software/gnutls/comparison.html yaSSL, dual-licensed (available as either GPL or commercial): http://www.yassl.com/ -- This post brought to you by: Ethan Grammatikidis, who follows directions like a butterfly.