[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-630562320 @mxnet-bot run ci [unix-cpu, website] This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-630468052 > Should the version dropdown been made consistent with the global version dropdown? I mean at least default to the same version number as used in the global version dropdown? > > > Not planned to add it to API pages while there is already a search bar on those pages. > > Should we replace that search bar? It seems your search implementation is superior? Good point, I will make the default version the same as the dropdown version. There is plan to improve the search bar in API sites, but that wasn't discussed in this project. For this project the objective is to add search for main site. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-630436025 > Thank you! > > Is it intended that the search queries entered while on the master version of the website yield results on 1.6 branch? I think it's better to be consistent > > What's the plan to add the search button on other parts of the page besides the index page? Thanks for the feedback! Is it intended that the search queries entered while on the master version of the website yield results on 1.6 branch? I think it's better to be consistent + The search UI is consistent across all pages. There is a version select drop down on the right side of the search input field. It is currently set to 1.6 by default and the search results are only relevant to that version. What's the plan to add the search button on other parts of the page besides the index page? + This feature is added to all information pages (tutorials, blogs, ecosystem, FAQ etc) on version master and 1.6 website. Not planned to add it to API pages while there is already a search bar on those pages. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-629832169 @mxnet-label-bot add [website] This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-629681651 @mxnet-bot run ci [unix-cpu] This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-629589289 @mxnet-bot run ci [unix-cpu] This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-629557829 @mxnet-bot run ci [unix-cpu] This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-629441115 > Very nice. Thank you @ys2843 This will be very helpful for the community. > Please update the release manager checklist for changing the defaults. > > Should we change anything with algolia (like default version) whenever there is a new release? > Also, for updating the search results, what is the process? What do you mean by improving the search result. > > Overall, this is really nice improvement for easily finding the content. Thank you. Thank you for reviewing. Will update the release manager checklist. > Should we change anything with algolia (like default version) whenever there is a new release? No, we don't need any change from algolia. We only need to update website code to set the default version. I will document this part. > Also, for updating the search results, what is the process? What do you mean by improving the search result. To make the search result more relevant and accurate, the process is to update the [configuration of DocSearch's web crawler](https://github.com/algolia/docsearch-configs/blob/master/configs/apache_mxnet.json). Currently the config is not ideal, and the best way to update and test the config is to run the web crawler locally. I have set up the crawler on an ec2 and run it with new config, and then make adjustment according to the result from the crawler. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-628990327 > Looks good. > > * Should we use offsite js files or host locally? > * Should the API key be in a js file are can it be in a config file so we keep this stuff in someplace central? > * Wonder if we could have the list of versions also in a config file and not in .js, so it is more obvious and easier to maintain... then other parts of the site like the install selector or other things that need the list of versions can share? > > Just some thoughts... otherwise this looks good to me. > Good work! Should we use offsite js files or host locally? + Based on the research, the CDN works pretty well in China. I will keep an eye on it, if there is problem loading this file in some country, I will switch to host the file locally. Should the API key be in a js file are can it be in a config file so we keep this stuff in someplace central? + Thanks for the suggestion. I will move API key to Jekyll config file Wonder if we could have the list of versions also in a config file and not in .js, so it is more obvious and easier to maintain... then other parts of the site like the install selector or other things that need the list of versions can share? + Great idea, let me remove it from JS and set these common variables as Jekyll global var This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
[GitHub] [incubator-mxnet] ys2843 commented on pull request #18288: Website global search feature
ys2843 commented on pull request #18288: URL: https://github.com/apache/incubator-mxnet/pull/18288#issuecomment-628867702 > Search feature looks great. > Nice to see we leveraging third-party to solve our search problem. > Q1. This feature is currently only for landing page right? > It isn't available on other pages right? > For e.g. not available in Python API > NDArray page right? > http://ec2-34-219-134-42.us-west-2.compute.amazonaws.com/api/python/docs/api/ndarray/index.html > > Can we document this point. And also add this functionality to rest of the pages in upcoming PRs? > > Q2 Is search suggestion pointing to the correct place? > When I search for copyto in 1.6 version on a laptop browser > It navigates to https://mxnet.apache.org/versions/1.6/api/python/docs/api/ndarray/ndarray.html?astype=#ndarray > > Isn't it supposed to go to the copyto operator specifically instead of the head of the div [ndarray] Thank you for reviewing! Good questions. Q1: Yes, it is available on all landing pages (that don't have a search bar currently), and not on any of the API docs pages. The reason is that while there is already a search bar on all API docs pages, in my opinion adding another one can confuse users. [Here is an example](https://pytorch.org/docs/stable/index.html), I think 1 search bar per page looks clear. Q2: Good point. This is related to improvement of search result, I am currently working on these issues. It is mainly about how we configure the web crawler to grab data from our website, and it does not require any code changes to MXNet website. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
