- 28 Oct, 2018 6 commits
-
-
Erik Huelsmann authored
-
Erik Huelsmann authored
-
Erik Huelsmann authored
-
Erik Huelsmann authored
Note that we no longer write the newsbox.html file, but instead put the content into the global context. Thus the global context needs to be passed to the expansion of the 'index.html' content. And if there, why not everywhere? (It'll open possibilities we haven't thought of yet...)
-
Erik Huelsmann authored
Note that the only remaining intermediate result being written to the 'input directory' content/, is the newsbox.html file due to it being a mustache partial. This commit 'publishes' the content to the global context instead and at the same time converts index.html to use a variable expansion. Thereby the need to write newsbox.html to disk has now been removed and so is the actual process.
-
Erik Huelsmann authored
Note that the news box generator used to receive just the 'main' tag's content whereas it now receives the entire HTML document. We need to expand the parser to extract the part it originally received and process that like we used to.
-
- 27 Oct, 2018 16 commits
-
-
Erik Huelsmann authored
-
Erik Huelsmann authored
Due to a missing 'and' before 'collect rel-page-path', the page's path was *always* added to the processed collection, instead of only when actually processed. Additionally, clean up output a bit.
-
Erik Huelsmann authored
Most output was already using relative paths; full paths have a tendency to overflow the terminal and thereby create confusing / hard-to-read output.
-
Erik Huelsmann authored
-
Erik Huelsmann authored
Now that we have pre-/post-processing and dependency tracking in place, we have all the components we need to stop generating 'news' before the rest of the site. The above is nice, because we no longer need to encode site structure knowledge into the site generator.
-
Erik Huelsmann authored
Note that we need to be able to declare a dependency on news.md from index.html, because we want to delay processing of index.html until news.md has been processed as we no longer want to process news twice. This commit is a step towards moving the generation of 'newsbox.html' into a post-processing step.
-
Erik Huelsmann authored
In order for the pre-/post-processor 'around-like' methods to work, they need access to the rendered page. Make sure they do by returning the content of the rendered page from the inner-most 'method' (=render-page).
-
Erik Huelsmann authored
Note that the LHTML preprocessor is now converted to an 'around method'-like concept which passes content and context from one around processor to another. Around processors may elect to pre- or post-process.
-
Erik Huelsmann authored
Note that the Markdown preprocessor is now converted to an 'around method'-like concept which passes content and context from one around processor to another. Around processors may elect to pre- or post-process.
-
Erik Huelsmann authored
This commit is one in a series which refactors how pages are being generated in order to retain the link between the original input document and the HTML document ultimately being generated.
-
Erik Huelsmann authored
This commit is one of a series which will result in retaining the original input filename being available when rendering the ultimate HTML page (which it currently isn't).
-
Erik Huelsmann authored
Note that in a follow-up, we'll need to add infrastructure to retain the original input filename as currently .md and .lisp files aren't linked to correctly (the link is to the resulting .html file instead).
-
Erik Huelsmann authored
-
Erik Huelsmann authored
-
Erik Huelsmann authored
-
Erik Huelsmann authored
Or really, build all subdirectories (including future subdirectory content)
-
- 26 Oct, 2018 6 commits
-
-
Erik Huelsmann authored
-
Erik Huelsmann authored
-
Erik Huelsmann authored
-
Erik Huelsmann authored
-
Erik Huelsmann authored
Remove still-born projects as per the announcement See merge request !72
-
Erik Huelsmann authored
-
- 25 Oct, 2018 1 commit
-
-
Administrator authored
-
- 22 Oct, 2018 5 commits
-
-
Mariano Montone authored
-
Erik Huelsmann authored
-
Dave Cooper authored
-
Dave Cooper authored
[metabang-bind](https://common-lisp.net/project/metabang-bind/user-guide.html#header3-21) supports the use of `_` as placeholder in binding list for variables which you want ignored. `nil` is also supposed to work, but appears not to in some cases. So this patch changes the use from `nil` to `_`.
-
Mariano Montone authored
-
- 21 Oct, 2018 4 commits
-
-
Dave Cooper authored
-
Dave Cooper authored
Because on some implementations, uiop:copy-file will not overwrite existing files, we explicitly remove existing destination files before copying with uiop:copy-file.
-
Dave Cooper authored
1. cl-markdown was crashing on its TOC processing on CL implementations (e.g. SBCL 1.4.*+) which have DESTRUCTURING-BIND treat non-toplevel () as a nested empty pattern, rather than just a shorthand for an ignored item as sometimes was being done in older, non-conforming code. The patch introduced by this commit fixes these non-comforming instances in CL-MARKDOWN. Note that CL-MARKDOWN is not using multiple-value-bind directly, but rather relies on the "metabang.bind" system - so maybe it can be fixed there. As noted in patches.lisp, a merge request has been lodged against cl-markdown. They are the same author as metabang.bind, so they can decide whether to accept this MR or fix it in metabang.bind instead. 2. On some implementations, the `process-static` function in `process.lisp` was spuriously picking up subdirectories in its `sources` list (where it expected just a list of source files). Now we are filtering out anything which looks like a directory (i.e. a non-file directory entry) by using `uiop:/filesystem:directory-exists-p`.
-
Erik Huelsmann authored
Markdown Closes #9 See merge request clo/cl-site!54
-
- 20 Oct, 2018 2 commits
-
-
Erik Huelsmann authored
Incorporates @mmontone's review remarks
-
Erik Huelsmann authored
-