search

clingen-data-model / allele

Documentation for data model of ClinGen
10 stars 2 forks source link

Broken links in resource model pages #165

Closed bpow closed 7 years ago

bpow commented 8 years ago

For instance, on http://datamodel.clinicalgenome.org/development/allele/resource/

The link to CanonicalAllele points to http://datamodel.clinicalgenome.org/allele/resource/canonical_allele/ (which gives a 404)

This is missing the "development" part of the path, it should point to http://datamodel.clinicalgenome.org/development/allele/resource/canonical_allele/

@tnavatar I think this has to do with the setup for building/publishing multiple versions of the documentation.

bpow commented 8 years ago

For some reason, altough link_to is being used in allele/resource/index.md.erb, and link_to is being set to use relative links (in config.rb), an absolute, rather than relative link is being used, and this breaks when using the 'development' or '0.1' prefix.

@tnavatar : any thoughts? It looks like you may have had to work around link_to issues before.

bpow commented 8 years ago

Assigning to @tnavatar, but let me know if you want me to look into this further.

bpow commented 8 years ago

You might be able to try adding set :http_prefix, ENV['MIDDLEMAN_PREFIX'] if ENV['MIDDLEMAN_PREFIX'] in the :build section of config.rb, then in the build script call MIDDLEMAN_PREFIX=$version bundle exec middleman build.

I'm not sure how that will interact with they way you had to handle relative links before this, though.

bpow commented 7 years ago

@tnavatar : we talked about adding an issue for this during the call today, but I see that I already had ;)

bpow commented 7 years ago

This looks to be fixed.

bpow commented 7 years ago

Apparently, this is actually not fixed, and we noticed again on the call today. Re-opening.

bpow commented 7 years ago

My best guess (after trying some fiddling) is that middleman's creation of relative links is not able to handle links to or filenames with '>' in them (that get url-escaped for publishing). Does this seem right to you, @tnavatar ?

If this is the case, can we just go ahead and rename those files to not have the full hgvs expression? Thoughts from @larrybabb or @cbizon ?

tnavatar commented 7 years ago

It’s possible, I haven’t encountered that case yet myself. > isn’t an acceptable character in RFC1738, so we should rename these files.

On May 30, 2017, at 1:04 PM, Bradford Powell notifications@github.com wrote:

My best guess (after trying some fiddling) is that middleman's creation of relative links is not able to handle links to or filenames with '>' in them (that get url-escaped for publishing). Does this seem right to you, @tnavatar https://github.com/tnavatar ?

If this is the case, can we just go ahead and rename those files to not have the full hgvs expression? Thoughts from @larrybabb https://github.com/larrybabb or @cbizon https://github.com/cbizon ?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/clingen-data-model/allele/issues/165#issuecomment-304944037, or mute the thread https://github.com/notifications/unsubscribe-auth/AAXWMPkew3CvjzQ4LtXZCc-Q_IHNbp7Bks5r_EwygaJpZM4HfLLR.

larrybabb commented 7 years ago

Sure, lets replace the > with an underscore (like c.500G_A). It isn't critical, just nice for the user to see a filename that communicates the variant that is contained within it. It does not need to be precise HGVS.

@tnavatar Feel free to change or let me know if you'd like me to.

bpow commented 7 years ago

I think I fixed this with ee1e894 by escaping the > with %3E in the uri references. (which is also what later versions of middleman would do)