<p>I don&#39;t think I&#39;ve read it anywhere, of any violence at <a href="https://mastodon.sdf.org/tags/handsoff" class="mention hashtag" rel="tag">#<span>handsoff</span></a> </p><p>That&#39;s remarkable</p><p>(You KNOW MAGA chuds would be squawking all over if there was even an minor altercation)</p><p>The turnout from older people, also remarkable - anyone with an IRA that counts on that to help pay their bills has just seen the funds cut by almost a third in the past few weeks - many of these folks no doubt voted <a href="https://mastodon.sdf.org/tags/Republican" class="mention hashtag" rel="tag">#<span>Republican</span></a> (but not any more, I&#39;ll wager)<br /> <br />The <a href="https://mastodon.sdf.org/tags/GOP" class="mention hashtag" rel="tag">#<span>GOP</span></a> has just lost a big segment of their reliable base</p>

<p><span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention">@<span>whitequark</span></a></span> re initial indexing speed: if you use ninja, there is a recently-merged PR (which should be in ninja 1.13) that may help with this if your project has multiple executable/lib targets:</p><p><a href="https://github.com/ninja-build/ninja/pull/2497" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://</span><span class="ellipsis">github.com/ninja-build/ninja/p</span><span class="invisible">ull/2497</span></a></p><p>so e.g. if your project builds an executable &#39;foo&#39; then: </p><p>ninja -t compdb-targets foo &gt; compile_commands.json</p><p>...gives you a compilation database with all commands required to build foo, and *only* those commands.</p>

<p><span class="h-card" translate="no"><a href="https://chaos.social/@ronya" class="u-url mention">@<span>ronya</span></a></span> <span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention">@<span>whitequark</span></a></span> The difference in initial indexing time comes from IntelliSense using a totally different parser to collect definitions across TUs, which is orders of magnitude faster but also much less accurate and much less detailed.</p>

<p><span class="h-card" translate="no"><a href="https://chaos.social/@ronya" class="u-url mention">@<span>ronya</span></a></span> <span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention">@<span>whitequark</span></a></span> IntelliSense is architecturally very similar to clangd here- they both use the build system to configure a full compiler frontend to run on some TU that contains the open file.</p>

<p>Today I had to read 7 abstracts and then the respective papers.</p><p>Can the authors please say in the first sentence what the paper is doing? Don&#39;t start by giving background, or saying why (you think) the problem is important.</p><p>Just please say right away what you have done. Then, if you feel it is really important, go on to give background justification of importance. Don&#39;t let me have to &quot;compute&quot; in order to determine what you have done by reading the abstract.</p><p>For all seven abstracts, I&#39;ve written a one- or, in some cases, two-sentence description of what they do and why we care.</p><p>An abstract where the main result/contribution is hidden in the middle of the second paragraph or three paragraphs is not an abstract!</p><p>(And don&#39;t start by saying, &quot;in this paper&quot;. Of course it is in this paper.)</p><p>1/</p>

<p><span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention">@<span>whitequark</span></a></span> AFAIU:</p><p>Intellisense:<br />- upside: works immediately on any code, as it’s generating the AST itself, similar to CodeQL.<br />- downside: limited ability to resolve things in precompiler conditionals.</p><p>Clangd:<br />- upside: knows exact types and dead code for your project based on the optional compile time settings you set. <br />- downside: basically useless until you have coaxed the current projects buildsystem to generate a compile_commands.json and built the project once. </p><p>Correct?</p>

<p><span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention">@<span>whitequark</span></a></span> what really bugs me about this is M$ literally came up with the idea of language servers. How come they <br />cannot write a proper implementation for C/C++?</p>

<p><span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention">@<span>whitequark</span></a></span> </p><p>but if i don&#39;t use pfp they see me as an elephant</p><p>that&#39;s even worse</p>

<p>if you&#39;re using VS Code (or editors based on it) and are frustrated with C/C++ IntelliSense being incredibly slow: ditch the Microsoft language server entirely and install clangd. clangd is actually fit for purpose, although the initial indexing is vastly slower</p>