<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">On 3/29/2018 1:08 AM, Richard Levitte
wrote:<br>
</div>
<blockquote type="cite"
cite="mid:20180329.100824.337723587531388798.levitte@openssl.org">
<pre wrap="">In message <a class="moz-txt-link-rfc2396E" href="mailto:1ce93d56-6fa4-1bae-d440-5ab843900e40@jordan.maileater.net"><1ce93d56-6fa4-1bae-d440-5ab843900e40@jordan.maileater.net></a> on Wed, 28 Mar 2018 17:10:40 -0700, Jordan Brown <a class="moz-txt-link-rfc2396E" href="mailto:openssl@jordan.maileater.net"><openssl@jordan.maileater.net></a> said:
openssl> Matt: Indeed, looks very promising. Now if only we were on
openssl> 1.1.1 :-(. I'm a little surprised that it doesn't read from a
openssl> BIO.
It's certainly possible to add such an API. As a matter of fact, we
do have that internally, specifically for PEM files... have a look in
1.1.1's crypto/include/internal/store_int.h. That's not the initial
intention with the API, though...</pre>
</blockquote>
<br>
To be clear: it doesn't bother me one way or the other. It just
seemed like the general design for "reading data from a stream" for
OpenSSL is to read from a BIO, rather than directly providing "read
from file", "read from memory buffer", et cetera. I was surprised
to see a new feature that didn't follow that pattern. I *do* need
"read from memory" for my application, but writing a temporary file
would not be a problem.<br>
<br>
<br>
<blockquote type="cite"
cite="mid:20180329.100824.337723587531388798.levitte@openssl.org">
<pre wrap="">Also, I can't quite shake the feeling that a BIO API would be a bit
shaky. Internally, the file: scheme loader opens all files in binary
mode, as it's designed to detect if the file is a PEM file or raw DER,
so the question remains, if we would open up a BIO STORE API, what are
th expectations? Will people open such files in binary mode at all
times? Should that be a content type agnostic interface (i.e. should
it detect if the file is PEM or raw DER), or should there be separate
functions for PEM and raw DER content?</pre>
</blockquote>
<br>
I would expect the behavior to be identical for a BIO interface and
any other - which, I guess, would mean that the BIO would need to be
in binary mode.<br>
<br>
<blockquote type="cite"
cite="mid:20180329.100824.337723587531388798.levitte@openssl.org">
<pre wrap="">Please note that for each question, we're getting further and further
away from the idea of having an interface where the caller doesn't
need to know much more than how to indicate where to load stuff from,
to an API that almost becomes a 1:1 mapping of PEM and d2i functions.
When we've come that far, what have we gained?
</pre>
</blockquote>
<br>
Agreed that for this interface the caller shouldn't need to know
what's being read.<br>
<pre class="moz-signature" cols="72">--
Jordan Brown, Oracle Solaris</pre>
</body>
</html>