ahelsing / tratihubis Goto Github PK

From gcf trac ticket 206:

Also attached the file omni_config-lnevers and omni_config-lnevers2.
Question [italics and underline]
The error reported by readyToLogin.py states: 
[indented and in italics]ERROR: No manifest found from describe at  https://www.emulab.net/protogeni/xmlrpc/am/2.0; review the logs.

[not indented or italics]What logs??

became:

___Question''__

The error reported by readyToLogin.py states: 

  ''ERROR: No manifest found from describe at https://www.emulab.net/protogeni/xmlrpc/am/2.0; review the logs._


What logs??

browser:sourcefilename should become a link to the source tree

["5 mins", "2 days", "5 days"]
Became:
[mins", "2 days", "5 days"]("5)

A commit hash that has a leading colon doesn't turn into a link:

This ticket is fixed in :a2174032783285923e11fd11da032e605c65ccad

See gcf ticket http://trac.gpolab.bbn.com/gcf/ticket/172

[03/Dec/2014:11:27:42 -0500] gets converted into a link

see protoch ticket 1250

On wiki pages, a crossout is text surrounded by ~. In Markdown, it is surrounded by ~~.

See protoch ticket 948

tratihubis creates labels for components if necessary, but not for anything else. Why not? It should.

Most ticket numbers are properly read as markdown links. So some things like (#12) seem to come out ok, while others like (#12). seem to fail. I can't tell why - is it the trailing period?

In trac, indenting the 1st line of a paragraph indents the whole thing.

After conversion, all that indentation is gone.

Se trac gcf 408 description

indents / spaces in a block quote from wiki have no indents in markdown.

see trac gcf tickets 320 and 323

If the Trac comment had a reply to someone else then it needs extra newlines at the end. See eg gcf trac 47

Reply to foo
> You said this
I say this

should be:

Reply to foo
> You said this

I say this

Input like this from gcf trac ticket 132:
*** ERROR: mapper: Reached run limit. Giving up. seed = 1335813330 Physical Graph: 4 Calculating shortest paths on switch fabric.

Becomes:
*** ERROR: mapper: Reached run limit. Giving up. seed = [1335813330](https://github.com/ahelsing/tickettest2/commit/1335813330) Physical Graph: 4 ...

If a wiki block quote uses syntax highlighting, like "#xml" to have it formatted as XML, that is not converted to the markdown equivalent. See trac ticket 323

In trac we use:

{{{
#xml
stuff here
}}}

In markdown that should become:

` ` `xml
` ` `

(without those extra spaces)

SSL error messages like [Errno 12345] are treated as wiki links into markdown link format like [12345](Errno)

ticket #603: another example of faulty translation of wiki link to markdown:

      "members_to_add" [{"SLICE_MEMBER" : "urn:publicid:IDN+ch.geni.net+user+gec17dem, "SLICE_ROLE" : "MEMBER"}],

became

      "members_to_add" [: "urn:publicid:IDN+ch.geni.net+user+gec17dem, "SLICE_ROLE" : "MEMBER"}]({"SLICE_MEMBER"),

Note though that if we avoided translation of things within block quotes, that would solve this.

In wiki, to get a link whose label is itself, you do [https://www.example.com]. Nothing is done to this during conversion, but the link is recognized. So you get a link inside square brackets.

See protoch ticket 801

It would be nice if ticketsToRender could take a range of VLAN tags, like 2-100,150,160-170.

For some sample code, look at VLANRange fromString and str methods: https://github.com/GENI-NSF/geni-tools/blob/develop/src/gcf/omnilib/stitch/VLANRange.py#L86

This rule: [r"^\s\d\.", r'#'] converts:

 1. list item one
 2. ...

Into

# list item one
 2. ....

which is clearly not right

See gcf trac ticket 146
Source had:

Omni knows about the following aggregates...
============================================
| Nickname | Version | URL                 |
============================================
  pg-gpo        2      http://www...../2.0

Translation:

Omni knows about the following aggregates...
========================================h4. 
| Nickname | Version | URL                 |
========================================
  pg-gpo        2      http://www...../2.0
  plc           1      http://www.....:12346

Another example of the problem that things in block quotes get converted.

See EG protoch ticket 1373

[Thu Apr 30 19:32:28 2015] [error] [client 128.89.254.216] D...

becomes

[Apr 30 19:32:28 2015](Thu) [error] [128.89.254.216](client) De

A clever regex could tell it that the first thing in the square bracket cannot be one of Thu/Fri/etc/client

I think text anywhere inside block quotes (which will be multiline) should not be translated. Making the regexes honor that would be tough though.

see gcf trac ticket 264:

ERROR:omni.protogeni:Cannot create a PG slice without a valid user credential: Your user certificate urn:publicid:IDN+pgeni.gpolab.bbn.com+user+inki has expired at 20130416163637Z. ProtoGENI users should log in to their account at the ProtoGENI website at

becomes

ERROR:omni.protogeni:Cannot create a PG slice without a valid user credential: Your user certificate urn:publicid:IDN+pgeni.gpolab.bbn.com+user+inki has expired at [20130416163637](https://github.com/ahelsing/tickettest2/commit/20130416163637)Z. ProtoGENI users should log in to 

code 65535: std::exception became code 65535:#### stdexception
see trac gcf 614

If there are 2 sets of empty single quotes, like '' then some stuff and then '', then the translator thinks you meant for that to be italic, and replaces each set of single quotes with an underscore.

See gcf trac ticket 421

ticket conversion is slow - relatively slow even when running without --really, although that makes it slower. Why?

• Try adding timestamps to the logger to see the problem.
• Do I need to cache the hub or repo instances?