Wiki Specifications Not Available All The Time

[brianking]

Problem

Specification pages that live on our legacy twiki instance are not available to all users all the time.

Example page: http://open-services.net/bin/view/Main/QmSpecificationV2

The error goes something like this:

Software error:

Insecure dependency in sysopen while running with -T switch at /usr/local/lib/perl5/site_perl/5.8.8/CGI/Session/Driver/file.pm line 107

Some technical details of the error are available here:

http://twiki.org/cgi-bin/view/Support/SID-02175

There does not seem to be a quick fix, and it is probably not worth investing in fixing this as we are migrating to new infrastructure soon for the site.

Possible Solutions

We need to port this data soon anyway with the new site coming, so here are a few possible solutions. Please contribute to discuss which is the most feasible, or if you see a better solution.

1. Move the documents to http://www.oasis-oslc.org/. As we are moving all specifications work to OASIS, this is the preferred option.

2. Move the documents to the new OSLC Mediawiki instance when it is setup

3. Export the documents to another format and make them available somewhere else on the site. Set up redirects from existing urls. There are some converters out there that can handle twiki format, such as Pandoc.

[berezovskyi]

All three options might involve URI changes. Another suggestion would be to crawl the snapshot of the wiki with, say, wget and serve the snapshot for the old URIs. We can then gradually update nginx (or whatever server you use) config to redirect old URIs to new pages if needed.

[berezovskyi]

cc @jadelkhoury

[berezovskyi]

@brianking @jamsden most of the specs can be cloned offline: https://aide.md.kth.se/open-services/bin/view/Main/WebHome.html

I looked on the server and this seems to be the command to run:

wget \
        --recursive \
        --level=1 \
        --no-clobber \
        --page-requisites \
        --html-extension \
        --convert-links \
        -e robots=off \
        --domains open-services.net \
http://open-services.net/bin/view/Main/OslcCore

[berezovskyi]

@brianking can we host the HTML crawled this way under https://github.com/OSLC/oslc-site-hugo/tree/master/static ?

[brianking]

@berezovskyi Yes we can put it where we like, and then just setup redirects.

Do you think /static is the best place (I am considering the url)? What about, for example, a sub-folder under?: https://github.com/OSLC/oslc-site-hugo/tree/master/content/specifications

The url would then be https://oepn-services.net/specifications/somefolder/file.html

[berezovskyi]

@brianking static is indeed a strange path for this. But I knew all files there will be deployed verbatim, I don't know how to get Hugo to deploy static assets from other directories. Ideally, we would have archive somewhere in the path to highlight that the content will not be updated. Plus I like the warning toolbar you have on oslc.co, we should use the same on this static HTML archive of the wiki.

[brianking]

So which do you think is better?

open-services.net/archive/something/file.html OR open-services.net/something/archiive/something/file.html

Let me know your preferred path specifically.

In Hugo, if the root folder already exists, add sub-folders and files there-in should 'just-work'. At least for mark-down because it will inherit the templates in themes/oslc/layouts/. For new root folders, we might need to add new templates. However, for HTML files it might work a little differently.

Maybe @jail can give some tips? Jail, we basically want to dump a load of HTML files in a folder somewhere and have them show up as-is when we run them through Hugo onto the live site.

[berezovskyi]

I would prefer this URL

http://open-services.net/wiki/requirements-management/

to appear under

http://open-services.net/archive/wiki/requirements-management/

But then for a year or so we would need to keep a 301 Moved Permanently redirect from the first one to another.

The actual specs are hosted under a URL like this

http://open-services.net/bin/view/Main/RmSpecificationV2?rev=57

Maybe we can cheat a bit here and redirect those to

http://open-services.net/archive/specs/RmSpecificationV2?rev=57

[berezovskyi]

cc @jamsden @jadelkhoury

[brianking]

I'm don't think having wiki in the url makes sense if there is no wiki.

[berezovskyi]

I see your point. The problem could be to relink all archived pages, this is why I was suggesting to preserve all pages on old URIs (we need to reserve them for redirects anyway).

Maybe I can just convert the content old specs into Markdown? Does the old wiki use Mediawiki markup? On Fri, 13 Oct 2017 at 16:58, Brian King notifications@github.com wrote:

I'm don't think having wiki in the url makes sense if there is no wiki.

— You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub https://github.com/OSLC/oslc-community/issues/3#issuecomment-336477677, or mute the thread https://github.com/notifications/unsubscribe-auth/AAD83uVZqcxnj04Txu_FfbEf8nfD9HjFks5sr3qWgaJpZM4ONMs1 .

--

–Andrew.

[brianking]

If we are talking about files under http://open-services.net/bin/view/Main/, it looks like you can pull all files in text format. They are in the folder 'data/Main' on the server.

We can also handle HTML format if needed.

[brianking]

2.0. Spec archive: http://oslc.co/specifications/archive/core-2.0/

Is that related to this issue? See: https://github.com/OSLC/oslc-site-hugo/issues/27#issuecomment-342456093

[berezovskyi]

Yes, it is. The main difference is that now I used the raw TWiki markup & pandoc to convert into Markdown.

For http://open-services.net/bin/view/Main/OslcCoreUiPreview, the source can be retrieved by clicking Raw View:

The conversion is described in https://github.com/OSLC/oslc-site-hugo/blob/master/content/specifications/archive/core-2.0/README.md; images need to be archived manually, and so do the references to them.

[brianking]

Are you able to take on the conversions for the other ones @berezovskyi ?