---
 src/gmnlm.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/src/gmnlm.c b/src/gmnlm.c
index 245ec85..7234954 100644
--- a/src/gmnlm.c
+++ b/src/gmnlm.c
@@ -43,6 +43,7 @@ struct browser {
 	struct Curl_URL *url;
 	struct link *links;
 	struct history *history;
+	bool sensitive, next_sensitive;
 	bool running;
[message trimmed]

Perhaps we could use <figure> and <figcaption>.

	<figure>
		<pre>preformatted text</pre>
  		<figcaption>alt text</figcaption>
	</figure>

Or maybe the aria-label attribute:

	<pre aria-label="alt-text">preformatted text</pre>

Now if only there was a way to distinguish between code and ASCII art.

---
The certificate notAfter timestamp in the known_hosts file was
incorrect because mktime expects time in local time, while
ASN1_TIME_to_tm returns time in GMT.

I split this patch into two.

 src/tofu.c | 21 ++++++++++++++++++++-
 1 file changed, 20 insertions(+), 1 deletion(-)

diff --git a/src/tofu.c b/src/tofu.c
index 3c161de..7cfd0c8 100644
--- a/src/tofu.c
+++ b/src/tofu.c
[message trimmed]

Certificates should not be considered expired until after notAfter has passed.
---
 src/tofu.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/tofu.c b/src/tofu.c
index 48a627f..3c161de 100644
--- a/src/tofu.c
+++ b/src/tofu.c
@@ -86,7 +86,8 @@ verify_callback(X509_STORE_CTX *ctx, void *data)
 	enum tofu_error error = TOFU_UNTRUSTED_CERT;
 	host = tofu->known_hosts;
 	while (host) {
-		if (host->expires < now) {
[message trimmed]

---
This patch fixes the notAfter timestamp in the known_hosts file.
The timestamp was incorrect because mktime expects time in local time,
while ASN1_TIME_to_tm returns time in GMT.

 src/tofu.c | 23 +++++++++++++++++++++--
 1 file changed, 21 insertions(+), 2 deletions(-)

diff --git a/src/tofu.c b/src/tofu.c
index af5d9f2..f18bc32 100644
--- a/src/tofu.c
+++ b/src/tofu.c
@@ -13,6 +13,24 @@
 #include "tofu.h"
[message trimmed]

I think that this functionality probably belongs in its own package.
It's out of scope for gmi.

This other package could implement an extensible gemtext renderer,
similar to existing markdown renderers (like github.com/yuin/goldmark).

On Fri Oct 16, 2020 at 7:12 AM EDT, alex wennerberg wrote:
> How do you feel about, for the HTML conversion, adding in-line media
> support? (e.g. what I do on https://alex.flounder.online/)

How would this work? Gemini doesn't support in-line media.

I think there might be a problem with section IDs. Take this example:

	# Title
	=> #Title

In a Gemini client, the link would make a new request to the current URL
with the fragment #Title.
But with this patch, it would be converted to the following HTML:

	<h1 id="Title">Title</h1>
	<p><a href="#Title">#Title</a></p>

And then the link would wrongly point to the header.

I added the function (Text).HTML to the gmi package so that other
projects can use it.

I'm not sure if section IDs are useful here, since Gemini does not allow
linking to sections.