TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

On the Shortcomings of Gemini Protocol

70 点作者 urlwolf将近 3 年前

11 条评论

geocar将近 3 年前
&gt; then AFAICT there&#x27;s no way to tell that the server is done sending the file.&quot;<p>This is incorrect. See <a href="https:&#x2F;&#x2F;gemini.circumlunar.space&#x2F;docs&#x2F;specification.html" rel="nofollow">https:&#x2F;&#x2F;gemini.circumlunar.space&#x2F;docs&#x2F;specification.html</a> section 4:<p><i>As per RFCs 5246 and 8446, Gemini servers MUST send a TLS `close_notify` prior to closing the connection after sending a complete response. This is essential to disambiguate completed responses from responses closed prematurely due to network error or attack.</i><p>&gt; There is no version number in the protocol, and the response layout was carefully constructed to make extending it hard<p>I think if you want HTTP, you should use HTTP.
评论 #31578078 未加载
efitz将近 3 年前
I love the idea behind Gemini but the people central to the project act like jack**es anytime you suggest something.<p>First time I saw Gemini on HN I got so excited and I emailed them and posted on the thread; I then subscribed to their email list server. The discussion there was along the lines of “yeah the discussion on HN was positive but just a bunch of feature requests from people ignorant of our tribal knowledge and we don’t like so meh”.<p>It would be much more productive if they were engaging and gently instructive, eg “that’s an interesting idea and we thought about it but ultimately we decided that it wasn’t aligned with our project; here’s a link to the email archive discussion”.<p>But instead they are completely dismissive.<p>Sure they’re free to act that way, but it is just self marginalization as people will stop trying to engage with them and ignore the project.
评论 #31581539 未加载
评论 #31578814 未加载
评论 #31581743 未加载
yakubin将近 3 年前
&quot;Web only for documents, not apps&quot; appeals to me. However, it&#x27;s not a question of just shaving off features. Some features of the web are extraneous for that purpose, and some are missing. For example, rendering maths on the web portably still requires you to make your page into an app. The defaults (wrt fonts e.g.) are poor. And while I&#x27;m at fonts and maths, you can&#x27;t assume the client has fonts appropriate for displaying maths, so now you need to bundle fonts with your page. The automatic hyphenation algorithm used in browsers is also visibly inferior to the one used in LaTeX.<p>And Gemini solves none of that. Maths is even harder with Gemini than with the web.
评论 #31581228 未加载
wdkrnls将近 3 年前
Scientists don&#x27;t need tables implemented in their markup. They need tables inside linked files that readers can actually use in data analysis without having to learn yet another complex technology. The browser could render linked CSV or SQLite files in a nice table widget for instance and people interested in the data could open it up in R and not have to mess around with parsing XML.
评论 #31578021 未加载
评论 #31579795 未加载
评论 #31580047 未加载
saidunlisted将近 3 年前
Regarding this:<p><pre><code> what it also doesn&#x27;t have is a way to say when the file has ended </code></pre> Is this not the stated purpose of TLS close-notify as described in the gemini specification? [1] Given that it follows a one-request per connection, this is a good signal for end-of-file, no?<p><pre><code> As per RFCs 5246 and 8446, Gemini servers MUST send a TLS `close_notify` prior to closing the connection after sending a complete response. This is essential to disambiguate completed responses from responses closed prematurely due to network error or attack. </code></pre> [1] Section 4: <a href="https:&#x2F;&#x2F;gemini.circumlunar.space&#x2F;docs&#x2F;specification.gmi" rel="nofollow">https:&#x2F;&#x2F;gemini.circumlunar.space&#x2F;docs&#x2F;specification.gmi</a>
mikewarot将近 3 年前
I just discovered Gemini a few days ago. I really was taken aback by the idea of no version number in the protocol. I kept reading, and tried out the 100ish line Python client, then the Lagrange client. It all gives me a nice fuzzy feeling.<p>I&#x27;m on a different quest, to create a Memex as originally envisioned. I&#x27;ll likely set up my own capsule and blog about things there, in parallel with my old blogspot blog.
评论 #31563568 未加载
评论 #31577982 未加载
评论 #31561995 未加载
carapace将近 3 年前
Before criticizing Gemini one should read the mailing list archives. I haven&#x27;t heard an objection yet that wasn&#x27;t raised and addressed already, typically months and months ago. You might not like Gemini, but it&#x27;s unlikely that you&#x27;ve got an original reason for it.<p>- - - -<p>I think of Gemini as like the Haiku form for poems: yes it&#x27;s highly restrictive, but the art is just in expressing yourself <i>though</i> the form.<p>People coming around and protesting that haikus aren&#x27;t sonnets, or that they aren&#x27;t suitable for writing plays, are missing the point, IMO.
midislack将近 3 年前
These aren&#x27;t valid criticisms of Gemini and seem to be front end webby complaints. No wonder &quot;flounder&quot; doesn&#x27;t want to reveal the identity of his or her likely non-stem soft science friend. Perhaps the friend is entirely made up.
icy将近 3 年前
What bums me the most is that I can’t wrap gemtext at my desired line length. I hate having to write “long lines” in my editor and prefer ‘gqip’ing them to 72 (that’s Vim for hard wrapping selected text). Terrible choice.
danShumway将近 3 年前
Gemini also has some issues with inline language changes if I remember correctly, which has some accessibility implications.<p>----<p><i>TLDR I think that Gemini has a lot more in common with platforms like PICO-8 than with other HTML replacements. It&#x27;s a niche community, and that makes me a lot more charitable towards &quot;nothing should change and we don&#x27;t care enough to address X technical concern&quot; takes than I used to be.</i><p>----<p>The thing that makes Gemini popular is also the thing that makes it easy to criticize: the spec doesn&#x27;t evolve and it&#x27;s not really seeing a ton of development, on purpose.<p>In other contexts, small concerns are something that would just get addressed later, but with Gemini they will never get addressed, so you kind of have to decide upfront whether or not you care about those concerns, because they&#x27;re never going away. On one hand, this makes Gemini great for some things (getting into client development, not needing to worry about the platform changing underneath you) and really bad for other things (being a general-purpose format for more than just hobby&#x2F;niche communities).<p>But to be fair (and despite how it gets characterized), I don&#x27;t actually believe that Gemini is <i>trying</i> to be anything other than a stable platform for niche communication, and I don&#x27;t mean that as an insult. It&#x27;s pretty obviously not trying to be the future of the web for normal people or a universal way that people write text online, or even a universal &quot;light&quot; format for communication, and while I do have criticisms of the protocol, I also don&#x27;t think that it has any real obligation to be the future of the web.<p>When weighed against some of the other proposals people have made to &quot;simplify&quot; the modern web (Amp, PDFs, Canvas, etc...), Gemini is noticeably better and noticeably more accessible, and actually has a real community of people using it. So it wins on all of those fronts. It&#x27;s just if you want it to be anything more than that, you&#x27;re going to be disappointed, because Gemini is deliberately designed not to be reactive to any needs that its authors didn&#x27;t already think about, which is pretty much game-over for wide adoption or longevity as a useful spec in the mainstream.<p>When people don&#x27;t understand that, it leads to fighting over &quot;is this feature actually important for a text-only web, do people <i>really</i> need this?&quot; But in my mind that&#x27;s the wrong way to think about Gemini&#x27;s downsides. There are many features that are important for a text-only web that Gemini doesn&#x27;t support because Gemini is not a foundation to replace HTML, and it&#x27;s less about whether Gemini supports every feature that would be good or even necessary for it to support if it was used in the mainstream, and more about whether it supports <i>just enough</i> features to be usable for a generally nice, semi-closed community. Most proposals beyond that turn out to be just bikeshedding.<p>It&#x27;s a little weird, but if you think of Gemini as more of a Tumblr&#x2F;Twitter alternative and less as an Amp competitor, then a lot of the refusal to evolve or address new concerns start to make a lot more sense. If you&#x27;re building a general-purpose text-based document format, then you have to care about things that Gemini doesn&#x27;t care about, and you can&#x27;t take an attitude of &quot;people&#x27;s needs matter less than keeping the spec small.&quot; But despite what the official website says, I don&#x27;t think that&#x27;s what Gemini actually is, and understanding that has made me feel a lot more charitable about the protocol than I used to be.<p>----<p>I will completely hypocritically say though that I do still think they should fix the `lang` problem since that&#x27;s just a general accessibility feature and is arguably the same or even less complexity than specifying languages in the header. But, hey, that&#x27;s just me proposing another spec change. :)
HidyBush将近 3 年前
Why would a scientist need to publish images and plots on the web? Just write a blog post which is basically an abstract of your article and then simply put a link to a gorgeous PDF containing the full article filled with gorgeous visualizations.
评论 #31579641 未加载
评论 #31579630 未加载
评论 #31581355 未加载
评论 #31587057 未加载