xmojmr / component.io Goto Github PK

Setup

Google Chrome
Set browser width to wider then 1150px
Sort by "freshness in days"
Reload page

Steps

1. Scroll down to make the table header fixed
2. Set browser width below 1150px and to make it switch to the shorter column list
   • expected behavior: table header resizes correctly, number of columns reduces to 8 and table header is fixed
3. Set browser width back above the 1150px
   • expected behavior: table header resizes correctly, number of columns grows to 10 and table header is fixed
4. Click the "stars" header to switch the table sorting from "freshness in days" to "stars"
   • expected behavior: header "stars" gets highlighted and header "freshness in days" gets normal and table data get sorted accordingly
   • actual behavior: header "stars" gets highlighted and below header "freshness in days" is small highlighted strip with text "days". There shoud be no additional visible strip

Server-side php code for Excel file generation will be needed. Exporting CSV data file (already now supported by the DataTables/TableTools extension is not very usable due to issues with utf8 encoding and delimiter..

This http://datatables.net/forums/discussion/11561/export-to-excel-willing-to-help old DataTables issue describes a way to go

Standard meaning (https://datatables.net/examples/basic_init/multi_col_sort.html) is ^ means A-Z ordering and v means Z-A ordering

e.g. tags for http://component.xmojmr.cz/#a=ndnhat display as comma delimited strings although the original keywords definition in the original component.json file is array is it should be

User story

As a developer I want to find stuff developed by "Mike Bostock" whatever is his current nick name in whatever community system I'm searching through. I just want to enter the name and get the results

Rationale

many developers are better known by their human name. e.g. "TJ Holowaychuk" is a well known name. While their GitHub nicknames ("visionmedia" in this case) are not well known outside the community. Using the human names would improve the human dimension of the search tool. It would be more visible who gets the credit ("..Credit is the primary currency of the free software world..") and much easier for newcomers to find friends

Implementation notes

This is not currently available in the component-crawler dataset but should not be difficult to get it (and cache it) through the GitHub API. Especially when the number of authors is quite small, definitelly smaller then GitHub Rate Limit 5000 requests per hour

Component Crawler is out of service for a few days(?) with it's further hosting and sponsorship future currently open.

Without this data source the search application has no data and no use.

It would be nice if such temporary data service outage could be handled by the search tool itself, with the use of it's own simple PHP-based GitHub crawler.

Using Miloslav Hůla's github-api PHP library was already discussed and it seems to be the easy way to reduce 3rd party dependencies

The problem is the clould filter window position: fixed and calculation of the window center in modal.js

This behavior is disturbing. If I use a bookmarkable url and share it with someone else and the someone else opens the url then a blinking caret is shown in the search box - without any explanation and without any clear sense

If the exact page length is not specified by the url then for desktops the default might be the current 100 records, while for mobiles ($('html').hasClass('touch')) some 10 or 20