New MySQL Tools and Features Enhance User Experience on Windows; Help to Lower Database Total Cost of Ownership by up to 90 Percent
For about a week, instead of hosting several hundred jQuery plugins and several thousand advertisements for laptop batteries, our plugins repository has been serving up a pretty pathetic message about spam and an allusion to some “new submission process.” This happened very suddenly, and we’re sorry to everyone who’s been inconvenienced. Please allow me a few minutes to explain what happened, where we’re headed, and how it impacts you. If you’re in a rush, here’s the short version.
Though the plugins site you’ve known
and loved was a valuable tool when it was first set up, it gradually became something of a white elephant for the project. While powerful distribution tools like GitHub and npm have come to the fore, we’ve been stuck in an aging, CMS-oriented paradigm that frustrated developers and consumers of plugins alike. Many people moved onto alternative sources for finding and vetting plugins. Furthermore, the sites’s original implementors and maintainers had since moved on from active involvement within the jQuery project. While the team faced a steady stream of complaints about usability and a general lack of features, the site itself faced a veritable barrage of (several flavors of) spam. There was your standard, keyword-rich SEO garbage, but there was also something slightly more insidious: the batch posting of under-documented, demo-free plugins with links to paid download sites by third parties trawling for affiliate cash. While this wasn’t explcitly disallowed, it led to a terrible, confusing experience for users and gave the site the distinct sense that all was not on the up and up.
The Best Laid Plans…
Sensitive to all these problems, the team began drawing up requirements for a complete overhaul. We knew we wanted to get out of the business of accepting uploads, serving downloads, and generally legislating the plugin “release” process on our own servers. With our official projects already on GitHub, we knew we wanted to leverage the many wheels they’d already invented for distribution, versioning, and facilitating open, collaborative development. Another requirement was to create a standard schema for authors to clearly delineate dependencies, like which version(s) of jQuery a plugin supports, as well as other plugins and CSS assets it requires to be functional. We also wanted to introduce some quality control, with both passive mechanisms like requiring GitHub and a manifest file for distribution, as well as active ones like user ratings. Finally, we wanted a clean slate; with stricter requirements for submission, there could be no mass import of all the old plugins. We made a lot of progress preparing specifications, but hampered by a lack of resources and a number of other projects, we never were able to get too deep into the implementation phase. After all, the old site “worked!”
…Often Go Awry
As the glut of spam grew worse and multiple reports started showing up on the jQuery Core bug tracker, I wanted to at least take some steps to wipe a bit of the spam and egg off our face. With newly provisioned access to the administrative tools on the site, I teamed up with the Drupal Views Bulk Operations module and set out to identify and delete spammers and their posts. Within a day or two, nearly 90% of the alleged “content” was gone from the site. I continued to monitor the situation over the ensuing days, and deleted spam as it came in. Unfortunately, I likely cast too wide of a net, and threw out several perfectly good babies with the bathwater. Even more unfortunately, I didn’t back up the database before I began this process. At this point, you can probably see where this story is headed. Later in the week, while I was attempting to delete four spam items, I was left completely horrified when the results of the operation reported that the remaining 10% — every single plugin remaining in the database — had been purged. All that remained was a year-old backup. Of course, I realize that this is a dreadful outcome, and I take full responsibility for it. If it helps, I am very receptive to hate mail and tweets reprimanding me for being irresponsible, unprofessional, or just stupid.
At this juncture, we were left with two choices.
- Keep on kicking the can down the road: Restore from the old backup, losing a lot of data anyway, and have authors go through the painstaking process of re-uploading their plugins, even though the site was slated to be deprecated entirely anyway.
- Use this glaring mistake as the impetus to hunker down, cleave from the past, and finally implement the site we’ve talked about for so long.
We’ve chosen the latter.
Nuts And Bolts
Over the past few days, we’ve started converting our plans into action, building out an infrastructure that’s backed by GitHub. There are two requirements for listing a plugin on the new site:
- A valid package.json file
We’ve followed the lead of CommonJS and npm and created a schema for specifying dependencies, delivery, and other metadata of jQuery plugins. While the format is largely similar to those other projects, we’ve had to make some minor tweaks to account for some plugin-specific details.
- At least one versioned release
This means having tagged your release point(s) with a valid semantic version number (semver) string.
We’ve pared down the submission and maintenance process to a single, one-time step: adding a post-receive hook to your plugin’s GitHub repository. Assuming your plugin meets the guidelines, a page will be created on the plugins site to present your usage and download information. We’ll keep track of new releases as you push them.
In The Interim
We recognize that the old site was still serving as an regular resource for a lot of people, especially newer jQuery users, who simply valued the existence of a central browsing location, despite its flaws. Though many experienced users had moved on to other sites, or relied on relationships with trusted authors and word of mouth, these can take time to develop. Until we’re able to launch the new system, we’re happy to direct you to several other directories and people who can help pick up the slack:
If, in the process of searching these or any other directories, you are directed back to the current site at plugins.jquery.com, typically a quick search for the author’s name and the plugin name will yield an alternative site where it was hosted. In addition, DailyJS just did a useful roundup of alternatives that goes into more detail.
Next Steps For Plugin Authors
If you’re a plugin developer who wants to make sure your plugins will be ready to go on day one (or even beforehand), you’ll want to make sure your plugins are up on GitHub, and then you’ll want to get started on creating your package.json files and making sure your versions are appropriately tagged. If you aren’t already familiar with Git and GitHub, then this is probably a very good time for you to take the plunge and get started. If you prefer another SCM system, you can look into setting up a mirror to git from svn or hg, or other project hosting sites, like bitbucket or gitorious. If you don’t develop your plugins as open source, or you don’t use any source control at all, we will not be able to accomodate you at this time.
How Can I Help?
Though the site is still a work in progress, you can track progress and even set up your own local development right now at github.com/jquery/plugins.jquery.com/. We’re still working on getting a public staging environment together.
Though the site is essentially “powered by” GitHub, there are obviously a number of moving parts in play. The site will actually be served from inside of WordPress, which is populated with the contents of the indexed plugins using a node.js tool. There are still many kinks to be ironed out, so if that sounds like a stack you want to jump on, please join us over at the repo. We’ll be using GitHub Issues for all feature requests, bugs, and discussion.
While we’ve put a great deal of thought into putting together our
package.json schema, it’s still a living document, and if you have questions or comments on why it is the way that it is, head on over to this issue.
In an ideal world, this certainly wouldn’t have happened exactly as it did. Sadly, it did. We hope you’ll accept our apologies for the many ways this transition might make your job harder, give you some extra work you weren’t expecting, or just plain leave a bad taste in your mouth. If you are able to forgive us, bear with us, and maybe even lend a hand, we’re confident we’ll be able to deliver the modern, useful plugins site the community deserves.
So that was a lot of words, here’s the takeaway.
- We’d been planning on replacing the original, spam-ridden plugin site for quite some time
- In the process of deleting on the spam, all the plugins were deleted and we didn’t have a recent backup
- Instead of burning cycles keeping the old site on life support, we decided to make a clean break and kick development on the new site into high gear.
- The new site is powered by GitHub and a package manifest for plugins
- Plugins from the old site were never going to be automatically imported into the new one.
- The repo is at github.com/jquery/plugins.jquery.com
- We are very sorry, but also very excited!
We’ve gotten a lot of feedback since last week’s announcement about the plugins site’s unfortunate tumble into oblivion, and I’d like to address a few of the most important concerns that have surfaced since.
“Could you make the old backup available for posterity?”
Yes. We can — and have. Over the weekend, we restored the most recent backup we had, and the original site is now living at archive.plugins.jquery.com; you should be able to browse through everything that’s there to your heart’s content. We also applied the most recent user information we had, so if you had an account on the old site at any point in the last year, it should still work. However, the site is closed to new user registrations. If you really need a new account, please get in touch with me personally and I can get that straightened out for you. We’ve also set up a redirect, so that if you should encounter any links to plugins.jquery.com in your browsing, you’ll (hopefully) end up at the corresponding page in the archive.
Just get a backup from the Wayback Machine!
While the Internet Archive has cached versions of content that was updated more recently than last October, we just don’t have the people-power to re-create the lost posts manually in the new archive site. If you have an account, you can feel free to add “new” or old plugins, or update existing ones, should you desire to. However, this archive will not be indexed by search engines.
If you hate CMS-es so much, what’s with WordPress?
We’re in the middle of a network-wide redesign, and WordPress offers us a valuable set of tools when it comes to theming, searching, and serving a group of sites. Our new motto, however, is pull requests, not passwords; we’re implementing theming, documentation, plugins, and more in such a way that contribution will not actually require an account on our CMS at all. As I outlined in the initial post, the plugin submission process will only involve adding a post-receive hook to your repository. In the event of a similar catastrophe, we’re made sure we’ll be able to replay the entire plugin contribution history and get the site back up to speed right away. Our goal is to leverage the WordPress features we find useful without it serving as a barrier to entry or as the canonical warehouse of content. If you are of the mind that WordPress is always a bad idea, no matter what, no matter how, you’re certainly entitled to that opinion, but at this point, it’s not particularly beneficial to the conversation.
Git(Hub) is hard
The new plugins site will serve as an index of plugins, with a simple “download” button right on each plugin’s page. You will not have to just browse around GitHub looking for jQuery plugins. If you don’t know git and only ever want to download jQuery plugins, you don’t have to learn it. However, if you want to submit plugins, you’ll have to be using some sort of source control that you can at least mirror in git. This is by design: it can be really easy to build a jQuery plugin, but that doesn’t mean it’s necessarily fit for public consumption. Requiring the use of source control and
package.json are passive mechanisms that will help ensure that plugins which proliferate are authored by developers who have met a reasonable baseline (and aren’t selling batteries). We’re only targeting GitHub support for launch, but we’d like to add support for other services as well. We are actively avoiding the use of GitHub-specific features that would force us to limit the site to GithHub users permanently.
It’s A Conspiracy!
Some have called into question the veracity of my account, and that’s understandable, given the timing and circumstances. But believe me, the last thing I wanted to do after spending a day manually pruning spam from the directory was turn around and cause a gigantic headache for thousands of people, including myself and my colleagues. I hope the re-launching of the last backup at least partially allays these concerns. Additionally, we’re starting off with GitHub simply because it has a very broad user base already, and it has been incredibly positive for us since we shifted to it for development of jQuery Core, UI, and Mobile.
Thanks again for bearing with us during this transition.
Vamos a ver como se puede importar un fichero de dump directamente a una base de datos desde la misma linea de comandos.
Esto surge como necesidad cuando tratamos de importar (como hacen la mayoría) un dump de una base de datos con phpMyAdmin y se fracasa en el intento. La mayoría de veces es debido a que el fichero es demasiado grande para que (debido a la configuración del servidor) phpMyAdmin pueda importarlo de forma correcta.
Veamos el comando que debemos utilizar y posteriormente explicaremos cada opción:
$ mysql -uadmin -ppassword -hlocalhost database < fichero_dump.sql
Describimos un poco las opciones, aunque viendo el comando se hace bastante fácil identificar que es cada cosa:
- -u es para especificar el usuario (el cual deberá tener permisos en la base de datos en la que tenemos que importar)
- -p es el password del usuario para acceder al servidor
- -h es el host donde queremos conectarnos, si no se especifica se considera que queremos conectar a localhost
Posteriormente a los comando, especificamos el nombre de la base de datos en la que queremos importar junto con el operador < que indicará la dirección en la que los datos viajaran (ya trataremos otro día como exportar desde linea de comando) y por último el nombre de fichero del dump queremos importar.
Con esto ya podemos importar en cualquier base de datos rápidamente nuestros datos.