[rtmpdump] Fwd: RTMPDump can no longer build against OpenSSL (versions after 1.0.2, supposedly)

[Howard Chu]

Gordon Steemson wrote:
> Apologies – for whatever reason, this message (alone, out of all others) ended up getting spamfiltered.  Good thing I check it every now and then.
> 
>> On Sep 11, 2023, at 9:51 AM, Howard Chu <hyc at highlandsun.com> wrote:
>>
>> Not sure whether we should patch anything here. A lot of the functions
>> we use are deprecated in newer versions of OpenSSL, and will eventually
>> disappear. But RTMPE will continue to use these mechanisms as-is. We
>> might just document that you must use OpenSSL older than 1.1 because the
>> necessary algorithms are unsupported in newer OpenSSL releases.
>>
>> Thoughts?
> 
> Sounds reasonable, but on the other hand, it didn't look like much needed to be changed to fix this – IIRC, two functions declare a local variable for that structure instead of a pointer to one.  I guess it is a balance between "is this worth putting any time into" and "does that specific version bump in OpenSSL actually matter in terms of improved security or whatever".

My build against OpenSSL 3 had a lot more errors than the two you mentioned.
If it had just been a couple line fix I would have just done it, but there's
more going on than just that.

As for the question you pose - as far as RTMPE goes, the answer is going to be
a resounding no. If anyone is using RTMPS, I guess keeping current with TLS
standards might be worthwhile, but again, the actual RTMP servers out there
are probably also EOL. No point in releasing new servers if the primary client,
the Flash player, has been dead for 2 years already.
> 
> Gordon S.
> 
>>> ---------- Forwarded message ---------
>>> Hi there,
>>>
>>> After considerable research, I have worked out that your file
>>> `handshake.h` declares an actual variable that gets mapped to one of
>>> OpenSSL's internal opaque types, even though all of them are obtained
>>> via constructors and destructors and only ever accessed by reference.
>>> This causes my compiler to go "WTF? How big is this thing supposed to
>>> be?" and immediately fail the build.  I'd submit a fix but I ran
>>> across this in the process of trying to bootstrap my system back to
>>> minimal functionality, and do not yet have a working git install,
>>> among other things.
>>
>> -- 
>>  -- Howard Chu
>>
>> _______________________________________________
>> rtmpdump mailing list
>> rtmpdump at mplayerhq.hu
>> https://lists.mplayerhq.hu/mailman/listinfo/rtmpdump
> _______________________________________________
> rtmpdump mailing list
> rtmpdump at mplayerhq.hu
> https://lists.mplayerhq.hu/mailman/listinfo/rtmpdump
> 


-- 
  -- Howard Chu
  CTO, Symas Corp.           http://www.symas.com
  Director, Highland Sun     http://highlandsun.com/hyc/
  Chief Architect, OpenLDAP  http://www.openldap.org/project/