A-Qute blog (windows)

Instalar Jekyll (y Ruby) en Windows

Tue, 29 Sep 2015 00:00:00 GMT

Pequeña guía para tener una instalación de GitHub Pages en Windows, para testear Jekyll antes de "pushear" los cambios. Recomendado cuando estas haciendo un theme de un blog o creando un sitio.

Instalar Ruby

Se necesita Ruby y Ruby Dev-kit para poder instalar los "gems".

Primero descargar ruby de Ruby Installer. Al instalar, checkear la opcion de "Add Ruby executables to your PATH".

Note

Al 2016-02-09, GitHub Pages usa la versión 2.1.7; la mas cercana listada es la 2.1.8. Podes descargar el instalador de Ruby 2.1.7 en los archivos cosa de asegurarte.

Del mismo sitio, descargar el Ruby DevKit correspondiente a la versión. La descarga es un archivo descomprimible .exe. Descomprimirlo en alguna carpeta sin espacios (por ejemplo: C:\Ruby\DevKit\).

Para "unir" el devkit e instalarlo, seguir los siguientes pasos:

con la línea de comandos ir a la carpeta (cd path\carpeta\)
Ejecutar ruby dk.rb init, esto auto-detecta las instalaciones de Ruby y añade a un archivo de configuración necesario para el próximo paso.
Ejecutar ruby dk.rb install para "unir" el DevKit.

Note

Si da error de "Comando no encontrado" verificar que en el PATH este como [path\a\ruby\]bin\ (notar barra invertida final)

~~Si vas a necesitar "Syntax Highlighting" y elegiste pygments, tenés que instalar Python (para Jekyll 2 o inferior).~~

Al 2016-02-09, GitHub Pages solo acepta Rouge para "Syntax Highlighting" (escrito en Ruby).

GitHub Pages local: instalación de Jekyll y complementos

En Windows no se puede instalar el gem de GitHub Pages por que tira error al querer instalar RDiscount, porque (según cita):

(RDiscount) It uses a POSIX function that is not available under Windows, even if you adjusted most of it.

(Luis Lavena en Super User)

Info

Al 2016-02-09, GitHub Pages solo acepta kramdown; desconozco si sigue tratando de descargar RDiscount (parece que si).

Por esto, es mejor instalar bundler (gem install bundler) y usar un gemfile (llamado "Gemfile") en donde se encuentra el proyecto, listando lo de arriba con las versiones que se usan:

    source 'https://rubygems.org'

    require "json"
    require "open-uri"
    #get versions specific to github
    versions = JSON.parse(open("https://pages.github.com/versions.json").read)

    gem "jekyll", versions["jekyll"]
    gem "jekyll-coffeescript", versions["jekyll-coffeescript"]
    gem "jekyll-sass-converter", versions["jekyll-sass-converter"]
    gem "kramdown", versions["kramdown"]
    #gem "maruku", versions["maruku"] #jekyll 3.0 uses kramdown
    #gem "rdiscount", versions["rdiscount"] #we cant install it in windows
    #gem "redcarpet", versions["redcarpet"] #no support in GH pages (2016-02-09)
    #gem "pygments.rb", versions["pygments.rb"] #no support in GH pages (2016-02-09)
    gem "rouge", versions["rouge"] #for code highlighting
    #gem "RedCloth", versions["RedCloth"] #textile, no support in GH pages (2016-02-09)
    gem "liquid", versions["liquid"]
    gem "jemoji", versions["jemoji"]
    gem "jekyll-mentions", versions["jekyll-mentions"]
    gem "jekyll-redirect-from", versions["jekyll-redirect-from"]
    gem "jekyll-sitemap", versions["jekyll-sitemap"]
    gem "jekyll-feed", versions["jekyll-feed"]
    gem "jekyll-seo-tag", versions["jekyll-seo-tag"]
    #gem "ruby", versions["ruby"] #avoid problems in windows
    gem "github-pages", versions["github-pages"]
    gem "html-pipeline", versions["html-pipeline"]
    gem "sass", versions["sass"]
    gem "safe_yaml", versions["safe_yaml"]

(idea del código de jamesjoshuahill)

Se instalan los gems (con bundler) yendo al directorio con el Gemfile y ejecutar bundle install.

Puede haber problemas con el certificado de https://rubygems.org. Si no se quiere probar, instalar (eliminado rdiscount, sin documentacion) los gems con el siguiente comando (al 2015-09-28):

gem install jekyll:3.0.2 --no-rdoc --no-ri jekyll-sass-converter:1.3.0 --no-rdoc --no-ri jekyll-textile-converter:0.1.0 --no-rdoc --no-ri kramdown:1.9.0 --no-rdoc --no-ri redcarpet:3.3.3 --no-rdoc --no-ri RedCloth:4.2.9 --no-rdoc --no-ri liquid:3.0.6 --no-rdoc --no-ri rouge:1.10.1 --no-rdoc --no-ri jemoji:0.5.1 --no-rdoc --no-ri jekyll-mentions:1.0.0 --no-rdoc --no-ri jekyll-redirect-from:0.9.1 --no-rdoc --no-ri jekyll-sitemap:0.10.0 --no-rdoc --no-ri jekyll-feed:0.3.1 --no-rdoc --no-ri jekyll-gist:1.4.0 --no-rdoc --no-ri jekyll-paginate:1.1.0 --no-rdoc --no-ri github-pages-health-check:0.6.0 --no-rdoc --no-ri jekyll-coffeescript:1.0.1 --no-rdoc --no-ri jekyll-seo-tag:0.1.4 --no-rdoc --no-ri ruby:2.1.7 --no-rdoc --no-ri github-pages:45 --no-rdoc --no-ri html-pipeline:2.3.0 --no-rdoc --no-ri sass:3.4.21 --no-rdoc --no-ri safe_yaml:1.0.4 --no-rdoc --no-ri

El error de certificado es un similar a:

[!] There was an error parsing `Gemfile`: SSL_connect returned=1 
errno=0 state=SSLv3 read server certificate B: certificate verify 
failed. Bundler cannot continue.

Se soluciona de la siguiente manera:

descargar el archivo cacert.pem y guardarlo en (por ejemplo ejemplo) C:\RailsInstaller\cacert.pem.
Settear el SSL_CERT_FILE usando la línea de comandos: set SSL_CERT_FILE=C:\RailsInstaller\cacert.pem. Se puede dejar "permanente" poniendolo en el PATH.

También se puede hacer por proyecto:

Guardar el archivo cacert.pem en el proyecto.

A list item with a code block:

ENV['SSL_CERT_FILE'] = File.expand_path(File.dirname(__FILE__)) + "/config/cacert.pem"

Correr Jekyll

Seguramente va a tirar un error de encoding, estilo:

jekyll 2.4.0 | Error:  incompatible character encodings: CP850 and UTF-8

para solucionarlo ejecutar en la línea de comandos chcp 65001 antes de ejecutar Jekyll.

En la carpeta donde se tiene los archivos:

Si se usó el fix de OpenSSL, hay que usar bundler: bundle exec jekyll serve
Si no: jekyll serve

Abrir en http://127.0.0.1:4000/

Fuentes:

Cambios --------- #. 2016-02-09 actualización a GitHub Pages usando Jekyll 3

Pelican to Nikola (and tutorial)

Sat, 11 Jan 2014 00:00:00 GMT

When I was thinking of moving my blog to this new domain, I thought of using Pandoc as the converter¹. I even created pandy for that. But then I run into the "problems" (ok, I was kind of lazy) of generating RSS, listing posts with certain tags or categories (which I ended up not using), the index and archive, idealizing of using plugins, etc. and gave up.

In the beginning, I would continue using Pelican as is it's easy and customizable, even though it doesn't have Pandoc² support. But later I would be lazy to move to another system (not that I blog much, but I like Pandoc's markdown).

Acrylamid

In my searching, I found Acrylamid which looks pretty good. It's like the programmer/hacker blog generator, where you can do many things, including using template syntax inside your post. How cool is that?. It also has incremental builds, which means that it won't regenerate the whole site/blog again after a little edit, but only what it was changed.

Other cool things include:

different meta tags: you can choose between MultiMarkdown (like Pelican), YAML or rst.
the draft status is a boolean. Yay!.
you can include template syntax inside your posts and you can even use the configuration variables!
automatic summary/intro
user custom acronyms and abbreviations in a text file.
the views things that looks cool that I couldn't understand first but then I got the hang of it (they are very customizable)
colored command line output. THANK YOU!
out of the box pingbacking to other blogs and twitter posting.
info about the blog: how many published, drafts, tag usage count. Also when was last compiled, cache size...
execute shell commands in the post: like listing a directory, include another file, etc.
serve the site with a command
post-processing filters

It doesn't have categories (but by using views seems like it's posible), automatic galleries creation (but you can use template syntax inside the post, so is covered) and the theme creation it's not so automatic: you must list all the files that the theme uses inside an __init.py file.

I really wanted to use Acrylamid, even without the missing bits, but it was kind of "complicated" to set up on Windows if you wanted to use python-magic (at least I have problems with it). I skipped that part but for some kind of reason I was getting errors with the locale setting. I tested it with "Linux locale" and "Windows locale" and both threw me errors.

Nikola

Then I remembered I once stumbled upon Nikola, originally created by an Argentinian guy, so yeah, I had to try it out :P.

It does have Pandoc support; but it's not as "feature rich" (or easy customizable) as Pelican, and the documentation it's not so "complete" if you will. There are a couple of things that are not documented and you must find it in the Google Groups, GitHub or looking in the code. And also, not so feature-cool like Acrylamid. However, I think it's on a good track and it's the newest of the three.

It has cool things too:

incremental builds (like Acryamid)
automatic gallery creation
categories (like Pelican)
easy themable (like Pelican)
serve the site with a command and it has "Live" re-rendering
retired posts (so no 404 errors)
redirections (in case you can't use .htaccess)
deploy commands (a bunch of commands to run, so you don't have to script it)
post-processing filters
internationalization support (the site/blog interface can be in many languages)
password protected posts/pages
"Schedule" posts/pages
Wide range of (3rd party) comment systems and AnnotateIt support.
reStructuredText Extensions (if you use that).
Support for Twitter cards
raw post/page support (to show the source in text)

The draft state is applied in the "tag" metadata instead of a separate one, and they are deployed by default; it has no color support for the command line (but I think it's in the talks) and the "show source of the post and copy it" it's enabled by default.

The moving

Install Nikola

You start first by getting Nikola and installing it. By default, it uses Mako as the template engine, but it has support for Jinja2. You can read about the dependencies in the documentation page and if you want, you can use pip install -r requirements.txt or requirements-full.txt for the whole thing.

If you are running Windows, you must download and install the compiled versions of lxml and pillow. You can check in their PyPI pages or you can go to http://www.lfd.uci.edu/~gohlke/pythonlibs/ and get them there. Remember to download and install according to your Python version and architecture (32 or 64 bits).

Initiate

To initiate or create an empty site with the default settings and folder structures, run either nikola init path/to/mysite or, if you want some demo content, nikola init --demo path/to/mysite.

Convert metadata

Pelican uses the Multimarkdown style of metadata while Nikola uses something similar: the same syntax but with .. in front.

An example would be:

Title: nice title! :D      # Pelican
.. title: nice title! :D   # Nikola

You can do that by hand or download my python script to take care of that. It also moves "draft" status to Nikola's tag, and Pelican's summary to Nikola's description (activated by default).

Note

Nikola doesn't have/see the "modified" meta key internally, but you can have it like a custom key. You just access it differently in the themes.

Translations

If you have posts or pages translations you must pay attention: Nikola handles the translations differently from Pelican. While in Pelican you only set the same slug or title in the posts, and have different file names (I think); Nikola relies on the file names.

The easiness of this depends on your setup, but you must end up with something like: nicefile.md for the post in default language, and nicefile.md.es (spanish content) for the translated file.

The cool thing of this is that you don't have to duplicate the meta tags. For example, you just create the "tags" meta key in the default language but don't specify it in the translated one. So when you update, you do only in the "master" or "parent" file and the "child" just copies it.

If you don't feel like having duplicated metadata between files, and after you renamed and converted them, you can run this python script to delete them. Remember this is optional (or at least it didn't break my setup).

Images paths

The syntax on this is also different but it can be easily done with search and replace. You can use whatever you want, like grep. If you are on Windows, I recommend grepWin if you want something with a GUI. This changes from setup to setup, but my settings were:

For markdown syntax:

Search (|filename|../../images/
Replace (/images/

And for HTML:

Search <img src="../../static/images/
Replace <img src="../images/

Links between posts

Again, this is totally different between the systems and it's going to be very easy or really difficult depending on your setup: Pelican uses the filenames for linking content while Nikola uses the slug of the post.

If you have the filename like the slug (or very similar) you can search and replace again. Like:

Search: $\|filename\|(.+)\.md$
Replace: (link://slug/$1)

I recommend doing it after the images, and you must check the links afterwards.

Plugins syntax

I didn't use many plugins that used syntax in my posts, just only one: interlinks. This was easilly done with search and replace.

Other Nikola stuff

Locale in Windows

If you are going to use the internationalization in Windows, you must configurate your LOCALES as (example):

LOCALES = {'en' : 'English_United States', 'es' : 'Spanish_Argentina'}

It doesn't matter your Windows language setting, I think it only accepts english locales (I tested it with Windows in Spanish and it worked).

Linking internally

If you want to link to another post, tag, archive, etc.:

gallery: link://gallery/gallery_name
tags index: link://tag_index
tag's page: link://tag/tagname
tag's RSS feed: link://tag_rss/tagname
blog's index page: link://index
index [NUMBER] page: link://index/NUMBER
per-year archive: link://archive/
posts of [YEAR] link://archive/YEAR
blog's feed: link://rss
other post: link://slug/the-nice-slug

Pandoc

As Pandoc uses the extension to recognize the input, you must configurate that extension in the COMPILERS option in the configuration file. If it's not found, it uses Markdown as default.

While it worked for me using markdown, it didn't exactly work with some others, like Mediawiki. I don't know why.

Posts folder and subfolders

Nikola preserves the folder structure of the posts in the subfolders. If you have file_g.txt inside folder_abc, that structure is kept in the output. It can be good if you want to use it as categories in the output. There is no way to configurate it in the configuration.

In Pelican you can have many subfolders and sort of categorize your posts as you wish, and the output is configurated (by default, by date) as you want.

Testing

Initially your posts go to a posts folder in your root and the pages in story. The images folder would go to files. You can configurate all of this in the configuration file.

After configurating your conf.py, which is very well commented, you can try it.

Build the site with nikola build. Then you can drag and drop pages to a browser or use the serve command: nikola serve and pointing your browser to http://127.0.0.1:8000

Because I hate opening an HTML block tag and the markdown inside is not parsed. I know, it's said in the markdown reference, but hey: it's 2014 and I need that. But I still love the admonition implementation in Python-Markdown ↩
Which now I see it looks pretty easy to implement. Oh well. ↩

Windows 7: Arreglar los iconos en Explorer

Thu, 12 Dec 2013 00:00:00 GMT

Con esta era de usar programitas varios para sincronizar, otros para control de código, y otros que no puedo recordar ahora; es necesario tener los overlay icons en Windows Explorer:

Personalmente se me desaparecían y, aunque limpiara la cache de los programas, nunca aparecían. Hay algunos que podes vivir sin ellos, pero otros no. Capaz instalaste un programa y te desplazó los de otro.

¿Pero cuál es el problema? Que los genios de Microsoft limitan la cantidad de iconos que se pueden usar a unos meros 11 iconos totales para el usuario. WHAT THE HELL.

Para arreglar el tema hay que pasar por el registro. Entrá al bonito regedit y navegá hasta HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer\ShellIconOverlayIdentifiers\. Antes de tocar mano, hacete un backup del .reg.

La idea es que de los 30 que debes tener, solo elegis los 11 mas importantes. ¿Cómo decidis la importancia? renombrandolos, porque Windows los elige por orden alfabético. Poneles números adelante o "z", "x", "y", esas letras que no se usan. Yo prioricé algunos de TortoiseSVN, un par de Dropbox, y el resto como sea el orden.

Una vez que estás feliz, reiniciá el explorer.exe: Administrador de tareas > matar explorer¹. Después lo inicias en Nueva tarea (o si queres desloguearte y loguearte, si queres reiniciar también, pero no hace falta).

Fuente: oliverpearmain: Dropbox and TortoiseSVN icons overlays in harmony

Kill, kill, kill! ↩

Templates en Zim Wiki

Fri, 15 Jul 2011 00:00:00 GMT

Zim es un programa de wiki de escritorio multiplataforma. En los foros decian que iban a presentar templates para las páginas por namespaces. Desconozco si las últimas versiones ya lo traen, pero en la versión para Windows no está.

Nota: es un template para toooodas las paginas nuevas, no por namespace.

Buscar donde se instaló el programa, e ir a: data\templates\wiki. Ahi hay un archivo de texto con el nombre Default.txt. Ese es el archivo a modificar para que cada vez que pongas "Crear una nueva pagina" salga con ese template.

En esta parte de la página oficial tenes un poco mas sobre el tema de los templates como variables y sintaxis.

XAMPP: Status Check Failure [3]

Tue, 31 May 2011 00:00:00 GMT

Un día, al actualizar XAMPP (el cliente de desarrollo que uso), empezó a tirar un "Status Check Failure [3]" al iniciar. Era medio molesto el mensaje por más que los servicios siguieran andando.

Busqué vaias soluciones, algunas decian que era el puerto ocupado y que lo usaba Skype. Skype no tengo. Otros decian que era el instalador, que no funcionaba con Windows 7 64 bits, pero aún descargando el archivo comprimido daba el mismo problema.

La solución fue algo asi:

Ir a las Variables del entorno (Equipo > propiedades > configuración avanzada)
Agregar al path a XAMPP (ejemplo: %SystemDrive%\xampp)
Buscan el acceso directo del xampp-control-panel.exe y van a sus propiedades.
Borren lo que dice en Iniciar en...

Fuente: Resolver el error explicado con imágenes en portugués

Lanzar aplicaciones desde Ejecutar

Sat, 18 Jul 2009 00:00:00 GMT

Hace un par de semanas/meses instalé nuevamente Launchy. Al poco tiempo (minutos) lo saqué. Me pareció no-muy-de-mi-gusto teniendo las aplicaciones que tenía corriendo en ese momento.

Pocos minutos después, me acordé de que había visto por algún lado hacer lo que hace Launchy (lanzador de aplicaciones) con la ventana de Ejecutar.

Hay que ir a Mi PC > propiedades > Sistema > Opciones avanzadas. Abajo hay un botón que dice Variables de entorno, entramos ahí.

Se abre una ventanita. Abajo, en Variables del Sistema, buscamos Path y damos click en Modificar

Sale otro Pop Up, y modificamos el valor de la variable. Lo único que hay que hacer es ir hasta el final, agregar ; (punto y coma) seguido de la ruta en donde se encuentran los shorcuts/atajos/accesos directos.

Aceptar, aceptar, aceptar. Listo! ahora Win + r, el nombre del shortcut .. et voilà!

UPDATE

Le dí otra oportunidad al Launchy. Me gustó, no consume tanto (o sí, dependiendo el gusto de cada uno).

Comments

Luxiano says:

July 19, 2009 at 1:53 pm

Yo también usaba Launchy hasta que lo saqué (también tengo poca RAM ¬¬)… y empecé a usar Ejecutar :D

PD: si, que tierna, no? jaja :P
A.Q. says:

July 19, 2009 at 7:28 pm

También tenes menos de 1 GB de RAM? Hagamos un club para que no se nos rían en la cara… :P
Luxiano says:

July 19, 2009 at 9:19 pm

Tengo 512 :( buaaa

El club para cuando quieras :P
A.Q. says:

July 20, 2009 at 2:27 pm

Yo tengo un poco mas 700 (grafica integrada)

Y… Ahora! pero ponele vos el nombre :P
Luxiano says:

July 20, 2009 at 2:43 pm

Yo tengo una placa de video con 64MB… :(

Nombre del club… subRAM? :P
A.Q. says:

July 20, 2009 at 3:16 pm

Luxiano y yo, A.Q., somos los dos primeros miembros del club subRAM. El único requisito es tener menos de 1GB de tal en estos tiempos (2009).

Si quieres unirte, no dudes en dejar un mensaje en este post.
Luxiano says:

July 20, 2009 at 9:49 pm

Un día de estos se hace la página :D
A.Q. says:

July 21, 2009 at 2:17 pm

:D