SqueakSSL plugin v0.2.0 (was: Re: Fixes to the SqueakSSL Plugin)

Hi Eliot,

Yes, the github version lacks that line. My pull request[1] has this fix, along with a few other fixes and improvments, but Tobias didn't like the image-sided SAN certificate validation scheme, so none of those have been merged. In the meanwhile I've started rewriting a few things, like logging, certificate serialization (so that it's accessible from the image), but I still haven't finished domain name validation.

Levente

[1] https://github.com/itsmeront/squeakssl/pull/3/files

On Fri, 24 Jul 2015, Eliot Miranda wrote:

Hi Tobias, Levente and Marcel,
    I'm just integrating your v0.2.0 sources into Cog and I see that in sqDestroySSL at line 139 in the Cog version of sqUnixOpenSSL.c there's the line

    if(ssl->serverName) free(ssl->serverName);

but that's missing from the 0.2.0 version.  It looks to me like the Cog version is correct.  Is that a storage leak in 0.2.0 or am I misreading the code (sqSetStringPropertySSL), and the line should not be
present?

On Wed, Apr 22, 2015 at 2:35 PM, <[hidden email]> wrote:

      Dear Smalltalkers

      as a follow up to the SqueakSSL changes sent around yesterday…

      On 21.04.2015, at 16:52, Tobias Pape <[hidden email]> wrote:
      > Dear Smalltalkers
      >
      > Starting with Levente Uzonyi's patches to the Linux version
      > of the SqueakSSL-Plugin, we (Marcel Taeumel and me) have
      > ported this to the OS X and Windows version as well.
      >
      > Find three binaries[1] that can be used instead of
      > the ones found on versions found on SqueakSSL's google code
      > page.
      > […]
      > [1]: http://forum.world.st/file/n4820846/squeakssl.zip


      … I'd like to thank Levente Uzonyi and Ron Teitelbaum for their
      input and providing infrastructure.

      I put together all files and have uploaded them to the new
      (in the googlecode-export sense) SqueakSSL gitHub project:

              https://github.com/itsmeront/squeakssl/

      And tagged a release as v0.2.0 with a Zip of all binaries (linux32/64,osx,windows)

              https://github.com/itsmeront/squeakssl/releases/tag/v0.2.0

      (scroll to bottom for the binaries zip: squeakssl_v0.2.0.zip)

      I hope this comes in handy for anyone who wants to play around with that.

      Best
              -Tobias




--
_,,,^..^,,,_
best, Eliot

Hi all,
On 25.07.2015, at 01:10, Eliot Miranda <[hidden email]> wrote:

> Hi Levente,
>
> On Fri, Jul 24, 2015 at 3:46 PM, Levente Uzonyi <[hidden email]> wrote:
> Hi Eliot,
>
>> Yes, the github version lacks that line. My pull request[1] has this fix, along with a few other fixes and improvments, but Tobias didn't like the image-sided SAN certificate validation scheme, so none of those have been merged. In the meanwhile I've started rewriting a few things, like logging, certificate serialization (so that it's accessible from the image), but I still haven't finished domain name validation.
>
> OK, I'll keep the version of sqUnixOpenSSL.c in Cog unchanged except for the use of SQSSL_VERSION for the version number.

Reading the Cog sources, sqUnixOpenSSL.c:422 reads
                case SQSSL_PROP_VERSION: return 1;
which should be
                case SQSSL_PROP_VERSION: return SQSSL_VERSION;

Just to make sure, the version SQSSL_VERSION should be 2.

>  Maybe you can integrate with the Cog version?  I like using %p instead of %lx,

done in c4cad6e81dac6e2475de3426dc613ec50226b230
(sorry, I mixed that with \r\n -> \n conversion, very sorry)


> and the use of (long)aSqInt variable is I think unnecessary.  Whether sqInt is int or long it=ll get passed as a long to printf, so the noise is unnecessary.

Yet, I'd get warnings:
if sqInt is an int,
        printf("%ld", aSqInt)
gives a format waring
        warning: format '%ld' expects argument of type 'long int', but argument 2 has type 'int'
so I would use %d like the Cog version does.
Yet, then, if sqInt is a long,
        printf("%d", aSqInt)
gives a format warning again:
        warning: format '%d' expects argument of type 'int', but argument 2 has type 'long int'

So I'd tend to not change the
        printf("%ld", (long)aSqInt);
as it works correctly and without warnings in both cases.



>
> Do you have installation instructions for LibreSSL and make flags for linux64 & linux32?
>

I try to put that together tomorrow, I managed to reach 1:30 am again...

Best
        -Tobias