"c++" tag is slugified as "c", causing false aliasing and breaking build with obtuse traceback

[mikemccracken]

Scanning posts.Traceback (most recent call last):
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/doit/doit_cmd.py", line 120, in run
    return command.parse_execute(args)
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/doit/cmd_base.py", line 78, in parse_execute
    return self.execute(params, args)
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/doit/cmd_base.py", line 245, in execute
    args)
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/nikola/__main__.py", line 220, in load_tasks
    self.nikola.gen_tasks('render_site', "Task", 'Group of tasks to render the site.'))
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/doit/loader.py", line 256, in generate_tasks
    for task_dict, x_doc in flat_generator(gen_result, gen_doc):
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/doit/loader.py", line 26, in flat_generator
    for item in gen:
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/nikola/nikola.py", line 1104, in gen_tasks
    for task in flatten(pluginInfo.plugin_object.gen_tasks()):
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/nikola/nikola.py", line 1098, in flatten
    for t in task:
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/nikola/plugins/task/archive.py", line 53, in gen_tasks
    self.site.scan_posts()
  File "/home/mmccrack/weblog/NIKOLA/local/lib/python2.7/site-packages/nikola/nikola.py", line 1205, in scan_posts
    other_tag = [k for k in self.posts_per_tag.keys() if k.lower() == tag.lower()][0]
IndexError: list index out of range

[Kwpolska]

Don’t differentiate between the two [solved].

Allowing every single character in existence might not be the easiest (nor sanest!) thing to do. We can try, though…

And the traceback itself to be handled in #1343.

[mikemccracken]

Sorry, I don't really understand your first comment. Is it solved in the
current master, or in some other code?
Or were you describing something I need to do to solve it? In which case,
that wasn't clear.

I worked around it by changing the tag to c-plus-plus, which is lame but
I'm not mad about it.

I won't argue for supporting every character, but if the slugified version
of a tag is not the same length as the current tag (or some other measure
of 'significantly different'), there should be a warning that shows them
both, IMO.

On Fri, Jun 27, 2014 at 8:04 AM, Chris “Kwpolska” Warrick <
notifications@github.com> wrote:

Don’t differentiate between the two [solved].

Allowing every single character in existence might not be the easiest (nor
sanest!) thing to do. We can try, though…

And the traceback itself to be handled in #1343
#1343.

—
Reply to this email directly or view it on GitHub
#1344 (comment).

[ralsina]

On 27/06/14 13:59, Michael McCracken wrote:

Sorry, I don't really understand your first comment. Is it solved in the
current master, or in some other code?
Or were you describing something I need to do to solve it? In which case,
that wasn't clear.

I worked around it by changing the tag to c-plus-plus, which is lame but
I'm not mad about it.

I won't argue for supporting every character, but if the slugified
version
of a tag is not the same length as the current tag (or some other measure
of 'significantly different'), there should be a warning that shows them
both, IMO.

What we need to do is notice if two tags, when slugified, match each
other and give a reasonable error. I thought there was code for it...

[Kwpolska]

What we need to do is notice if two tags, when slugified, match each other and give a reasonable error. I thought there was code for it...

There is — and it’s now broken (see #1343).

Sorry, I don't really understand your first comment. Is it solved in the current master, or in some other code?
Or were you describing something I need to do to solve it? In which case, that wasn't clear.

My “solution” was: nuke one of the tags.

[Kwpolska]

A “real” solution would involve making + a valid character in a slug. Might not be a good idea.

[ralsina]

Yes, there's two problems.

1. Do we allow "+" in slugs?
2. Let's give a better error when slugs collapse tags.

[Kwpolska]

Problem 2 fixed in #1343.

[ralsina]

Since "+" is a valid character in URL paths, I see no reason not to allow it in slugs. So, allowing it in upcoming branch.

[Kwpolska]

There are many more characters we could allow, why not try to do them too?

Chris “Kwpolska” Warrick http://chriswarrick.com/
Sent from my SGS3.

[ralsina]

Usually slugs are defined as letter+number+- ... I added + because there was a good reason, other characters need their own reason, I guess? What characters are we talking about?

[Kwpolska]

It's hard to think of a good reason right now. Then again, going too far
would pretty much invalidate USE_SLUGIFY.

Chris “Kwpolska” Warrick http://chriswarrick.com/
Sent from my SGS3.