💾 Archived View for gemi.dev › gemini-mailing-list › 000340.gmi captured on 2023-11-04 at 12:42:40. Gemini links have been rewritten to link to archived content
-=-=-=-=-=-=-
Hi all, I've been working on this in my spare time for the last week or two, still lots to do and it's buggy in places (especially tabs, and some pages aren't rendering (that _did_ work previously)) but it's shaping up nicely. Download, more info, links to source etc: https://www.oppenlab.net/pr/tva/
Nice! Works well for me! On 24 August 2020 14:41:26 BST, "?ppen" <tva at fastmail.se> wrote: >Hi all, I've been working on this in my spare time for the last week or >two, still lots to do and it's buggy in places (especially tabs, and >some pages aren't rendering (that _did_ work previously)) but it's >shaping up nicely. > >Download, more info, links to source etc: >https://www.oppenlab.net/pr/tva/ -- Sent from my Android device with K-9 Mail.
Hi, I started an ?awesome-gemini? list on GitHub to spread the idea of gemini and added your Android client as well: https://github.com/kr1sp1n/awesome-gemini <https://github.com/kr1sp1n/awesome-gemini> Please contribute if you wanna add something. Have a great day, Krispin > On 24. Aug 2020, at 15:41, ?ppen <tva at fastmail.se> wrote: > > Hi all, I've been working on this in my spare time for the last week or two, still lots to do and it's buggy in places (especially tabs, and some pages aren't rendering (that _did_ work previously)) but it's shaping up nicely. > > Download, more info, links to source etc: https://www.oppenlab.net/pr/tva/
This is looking really nice, aesthetics are great and it's really fast.
Curiously, I noticed that it fails when fetching many CGI-generated URLs,
all mine yield a blank screen, other sites have missing data.
I did a little checking, and it seems that (at least some) servers don't
properly signal their intention to close the TLS session when sending the
output of CGI scripts.
I wonder if your client picks up this error and abandons the response
(which in fact was received correctly)? If so, perhaps it could instead
flush any input buffers and keep the response.
I have put a few notes below in case it's of use to you.
Kevin
---------------------------------------
# Example URL that fails in Tv?: gemini://gemini.susa.net/cgi-bin/twitter
# Response is received, but with an error.
kevin at kakapo:~/gemini/content$ echo -ne "gemini://
gemini.susa.net/cgi-bin/twitter\r\n"|gnutls-cli --no-ca-verification
--port=1965 gemini.susa.net
Processed 151 CA certificate(s).
Resolving 'gemini.susa.net:1965'...
Connecting to '78.46.186.193:1965'...
- Certificate type: X.509
- Got a certificate list of 1 certificates.
- Certificate[0] info:
- subject `CN=*.susa.net', issuer `CN=*.susa.net', serial
0x00af7f3177de559661, RSA key 4096 bits, signed using RSA-SHA256, activated
`2020-07-10 21:47:06 UTC', expires `2030-07-08 21:47:06 UTC',
pin-sha256="kUjkK2Oa2yhHsyFHokB3df7DqbTxgIwzah2bLzZFAjw="
Public Key ID:
sha1:c6b2cce3070eba096a2ddede1b65d8684bb04ff1
sha256:9148e42b639adb2847b32147a2407775fec3a9b4f1808c336a1d9b2f3645023c
Public Key PIN:
pin-sha256:kUjkK2Oa2yhHsyFHokB3df7DqbTxgIwzah2bLzZFAjw=
- Successfully sent 0 certificate(s) to server.
- Description: (TLS1.2-X.509)-(ECDHE-X25519)-(RSA-SHA512)-(AES-256-GCM)
- Session ID:
10:69:F0:70:F3:B6:9B:B2:9E:DA:A3:7E:02:99:B4:86:BE:54:9E:94:FB:30:1F:66:C6:
B8:61:20:88:21:F8:FF
- Options: extended master secret, safe renegotiation,
- Handshake was completed
- Simple Client Mode:
20 text/gemini
#The Twitter Oracle spake thus, the words spoken were:
```
Fr u hoes WEAK is u serious??
```
Just a tiny sample of what you're missing out on - refresh for more!***
Fatal error: Error in the pull function.
Thanks Kevin, and yes, something is afoot with some addresses. The client torture test at gemini://gemini.conman.org/test/torture/ used to render fine, but in later builds it's failing. I'll investigate for the next release. On Tue, 25 Aug 2020, at 12:16 PM, Kevin Sangeelee wrote: > This is looking really nice, aesthetics are great and it's really fast. > > Curiously, I noticed that it fails when fetching many CGI-generated > URLs, all mine yield a blank screen, other sites have missing data. > > I did a little checking, and it seems that (at least some) servers > don't properly signal their intention to close the TLS session when > sending the output of CGI scripts. > > I wonder if your client picks up this error and abandons the response > (which in fact was received correctly)? If so, perhaps it could instead > flush any input buffers and keep the response. > > I have put a few notes below in case it's of use to you. > > Kevin > > --------------------------------------- > # Example URL that fails in Tv?: > gemini://gemini.susa.net/cgi-bin/twitter > # Response is received, but with an error. > kevin at kakapo:~/gemini/content$ echo -ne > "gemini://gemini.susa.net/cgi-bin/twitter\r\n > <http://gemini.susa.net/cgi-bin/twitter/r/n>"|gnutls-cli > --no-ca-verification --port=1965 gemini.susa.net > Processed 151 CA certificate(s). > Resolving 'gemini.susa.net:1965'... > Connecting to '78.46.186.193:1965'... > - Certificate type: X.509 > - Got a certificate list of 1 certificates. > - Certificate[0] info: > - subject `CN=*.susa.net', issuer `CN=*.susa.net', serial > 0x00af7f3177de559661, RSA key 4096 bits, signed using RSA-SHA256, > activated `2020-07-10 21:47:06 UTC', expires `2030-07-08 21:47:06 UTC', > pin-sha256="kUjkK2Oa2yhHsyFHokB3df7DqbTxgIwzah2bLzZFAjw=" > Public Key ID: > sha1:c6b2cce3070eba096a2ddede1b65d8684bb04ff1 > > sha256:9148e42b639adb2847b32147a2407775fec3a9b4f1808c336a1d9b2f3645023c > Public Key PIN: > pin-sha256:kUjkK2Oa2yhHsyFHokB3df7DqbTxgIwzah2bLzZFAjw= > > - Successfully sent 0 certificate(s) to server. > - Description: (TLS1.2-X.509)-(ECDHE-X25519)-(RSA-SHA512)-(AES-256-GCM) > - Session ID: > 10:69:F0:70:F3:B6:9B:B2:9E:DA:A3:7E:02:99:B4:86:BE:54:9E:94:FB:30:1F:66:C 6:B8:61:20:88:21:F8:FF > - Options: extended master secret, safe renegotiation, > - Handshake was completed > > - Simple Client Mode: > > 20 text/gemini > #The Twitter Oracle spake thus, the words spoken were: > > ``` > Fr u hoes WEAK is u serious?? > ``` > > Just a tiny sample of what you're missing out on - refresh for more!*** > Fatal error: Error in the pull function. > *** Server has terminated the connection abnormally. > > > On Mon, 24 Aug 2020 at 14:42, ?ppen <tva at fastmail.se> wrote: > > Hi all, I've been working on this in my spare time for the last week or two, still lots to do and it's buggy in places (especially tabs, and some pages aren't rendering (that _did_ work previously)) but it's shaping up nicely. > > > > Download, more info, links to source etc: https://www.oppenlab.net/pr/tva/
On 2020-08-24 16:41, ?ppen wrote: > Hi all, I've been working on this in my spare time for the last week > or two, still lots to do and it's buggy in places (especially tabs, > and some pages aren't rendering (that _did_ work previously)) but it's > shaping up nicely. > > Download, more info, links to source etc: > https://www.oppenlab.net/pr/tva/ Great android client! Looking really neat.
It was thus said that the Great ?ppen once stated: > Thanks Kevin, and yes, something is afoot with some addresses. The client > torture test at gemini://gemini.conman.org/test/torture/ used to render > fine, but in later builds it's failing. I'll investigate for the next > release. Is it all the tests, or just some of them? I can investigate from my side as well. -spc
I've solved it. For some reason creating a second buffered reader from the socket was causing the socket to close - I'm now just reusing the same reader for content that checks the header. Thanks all for the feedback - I'll get a new release out in the next few days with the fixes. On Tue, 25 Aug 2020, at 9:19 PM, Sean Conner wrote: > It was thus said that the Great ?ppen once stated: > > Thanks Kevin, and yes, something is afoot with some addresses. The client > > torture test at gemini://gemini.conman.org/test/torture/ used to render > > fine, but in later builds it's failing. I'll investigate for the next > > release. > > Is it all the tests, or just some of them? I can investigate from my side > as well. > > -spc > >
?ppen <tva at fastmail.se> writes: > Hi all, I've been working on this in my spare time for the last week or two, Hey, this looks really great, visually/design-wise. I noticed that it supports Android system dark/light themes. The light theme is, IMO, a
Jason McBrayer <jmcbray at carcosa.net> writes: > Hey, this looks really great, visually/design-wise ... The light > theme is, IMO, a *little* low-contrast, though it looks nice. Agreed on both points - i like the overall design, but currently find the the light theme too low-contrast. At any rate, i certainly appreciate the work so far! A technical point which might already have been noted: the Gemini speculative spec page is truncated for me. The heading "5.4.3 Preformatting toggle lines" is displayed, but the only text after that is a preformatted block containing the text "Preformatting toggle lines can be thought of as analogous to <pre> and </pre> tags in HTML." i'm guessing this might be connected with the presence of "```" in the first para of that section? Alexis.
Thanks both. Yes, light theme could do with a little attention. My device is permanently in dark mode so it's been neglected and light theme uses the Android defaults. Jason; someone else alerted me to the nested ``` issue just yesterday when some ascii art they had contained it. Simple fix will be in next release. Next beta will be sometime in next few days. Lot's of small bug fixes, and an in-app MP3 player (you click a file, it plays, you can carry on browsing). I've also removed tabs, they didn't help the browsing experience and were buggy anyway. If I have time I'll incorporate the GUS trackback service so you can easily see which pages link where. That'll be the last release for a while while I concentrate on something that is needed: bookmarks. I also need to check GUS for the most common filetypes and add support for those (even if they're just passed to the OS to handle (eg. .pdf)). I'm also thinking about a future aggregation feature, I'll do some research before clogging up this list but I think it should be possible for clients to detect a Gemtext document with a series of dates in and offer a 'follow' button. I keep finding interesting Gemlogs then losing them again, bookmarks will help with that, but an aggregated timeline would be a fantastic feature (I know Solderpunk has done some work with Atom feeds for Gemini but if it feasible I like the zero setup approach). On Thu, 3 Sep 2020, at 5:59 AM, Alexis wrote: > > Jason McBrayer <jmcbray at carcosa.net> writes: > > > Hey, this looks really great, visually/design-wise ... The light > > theme is, IMO, a *little* low-contrast, though it looks nice. > > Agreed on both points - i like the overall design, but currently > find the the light theme too low-contrast. At any rate, i > certainly appreciate the work so far! > > A technical point which might already have been noted: the Gemini > speculative spec page is truncated for me. The heading "5.4.3 > Preformatting toggle lines" is displayed, but the only text after > that is a preformatted block containing the text "Preformatting > toggle lines can be thought of as analogous to <pre> and </pre> > tags in HTML." i'm guessing this might be connected with the > presence of "```" in the first para of that section? > > > Alexis. >
Adding client certificates would be nice too, unless there's a system setting I'm missing. On Thu, Sep 3, 2020 at 6:23 AM ?ppen <tva at fastmail.se> wrote: > Thanks both. > > Yes, light theme could do with a little attention. My device is > permanently in dark mode so it's been neglected and light theme uses the > Android defaults. > > Jason; someone else alerted me to the nested ``` issue just yesterday when > some ascii art they had contained it. Simple fix will be in next release. > > Next beta will be sometime in next few days. Lot's of small bug fixes, and > an in-app MP3 player (you click a file, it plays, you can carry on > browsing). I've also removed tabs, they didn't help the browsing experience > and were buggy anyway. If I have time I'll incorporate the GUS trackback > service so you can easily see which pages link where. > > That'll be the last release for a while while I concentrate on something > that is needed: bookmarks. > > I also need to check GUS for the most common filetypes and add support for > those (even if they're just passed to the OS to handle (eg. .pdf)). > > I'm also thinking about a future aggregation feature, I'll do some > research before clogging up this list but I think it should be possible for > clients to detect a Gemtext document with a series of dates in and offer a > 'follow' button. I keep finding interesting Gemlogs then losing them again, > bookmarks will help with that, but an aggregated timeline would be a > fantastic feature (I know Solderpunk has done some work with Atom feeds for > Gemini but if it feasible I like the zero setup approach). > > On Thu, 3 Sep 2020, at 5:59 AM, Alexis wrote: > > > > Jason McBrayer <jmcbray at carcosa.net> writes: > > > > > Hey, this looks really great, visually/design-wise ... The light > > > theme is, IMO, a *little* low-contrast, though it looks nice. > > > > Agreed on both points - i like the overall design, but currently > > find the the light theme too low-contrast. At any rate, i > > certainly appreciate the work so far! > > > > A technical point which might already have been noted: the Gemini > > speculative spec page is truncated for me. The heading "5.4.3 > > Preformatting toggle lines" is displayed, but the only text after > > that is a preformatted block containing the text "Preformatting > > toggle lines can be thought of as analogous to <pre> and </pre> > > tags in HTML." i'm guessing this might be connected with the > > presence of "```" in the first para of that section? > > > > > > Alexis. > > > -- ? <https://www.google.com/teapot>
> Adding client certificates would be nice too, unless there's a system setting I'm missing That'll come as it matures. It's currently still using a basic network class derived from another (abandoned I think) proof-of-concept Android project. On Thu, 3 Sep 2020, at 3:48 PM, A. E. Spencer-Reed wrote: > Adding client certificates would be nice too, unless there's a system setting I'm missing. > > On Thu, Sep 3, 2020 at 6:23 AM ?ppen <tva at fastmail.se> wrote: > > Thanks both. > > > > Yes, light theme could do with a little attention. My device is permanently in dark mode so it's been neglected and light theme uses the Android defaults. > > > > Jason; someone else alerted me to the nested ``` issue just yesterday when some ascii art they had contained it. Simple fix will be in next release. > > > > Next beta will be sometime in next few days. Lot's of small bug fixes, and an in-app MP3 player (you click a file, it plays, you can carry on browsing). I've also removed tabs, they didn't help the browsing experience and were buggy anyway. If I have time I'll incorporate the GUS trackback service so you can easily see which pages link where. > > > > That'll be the last release for a while while I concentrate on something that is needed: bookmarks. > > > > I also need to check GUS for the most common filetypes and add support for those (even if they're just passed to the OS to handle (eg. .pdf)). > > > > I'm also thinking about a future aggregation feature, I'll do some research before clogging up this list but I think it should be possible for clients to detect a Gemtext document with a series of dates in and offer a 'follow' button. I keep finding interesting Gemlogs then losing them again, bookmarks will help with that, but an aggregated timeline would be a fantastic feature (I know Solderpunk has done some work with Atom feeds for Gemini but if it feasible I like the zero setup approach). > > > > On Thu, 3 Sep 2020, at 5:59 AM, Alexis wrote: > > > > > > Jason McBrayer <jmcbray at carcosa.net> writes: > > > > > > > Hey, this looks really great, visually/design-wise ... The light > > > > theme is, IMO, a *little* low-contrast, though it looks nice. > > > > > > Agreed on both points - i like the overall design, but currently > > > find the the light theme too low-contrast. At any rate, i > > > certainly appreciate the work so far! > > > > > > A technical point which might already have been noted: the Gemini > > > speculative spec page is truncated for me. The heading "5.4.3 > > > Preformatting toggle lines" is displayed, but the only text after > > > that is a preformatted block containing the text "Preformatting > > > toggle lines can be thought of as analogous to <pre> and </pre> > > > tags in HTML." i'm guessing this might be connected with the > > > presence of "```" in the first para of that section? > > > > > > > > > Alexis. > > > > > > -- > ? <https://www.google.com/teapot>
---