[blfs-support] WebKitGTK+2.12.4 unneeded sed?

Ken Moffat zarniwhoop at ntlworld.com
Sat Sep 17 09:20:02 PDT 2016


On Sat, Sep 17, 2016 at 01:31:14AM -0700, Craig Jackson wrote:
> On Sat, Sep 17, 2016 at 12:38 AM, Chris Staub <chris at beaker67.com> wrote:
> > On 09/17/2016 02:54 AM, Craig Jackson wrote:
> >>
> >> I understand a catchall, but it actually errors out as written in the
> >> book since the commands are chained with &&, and without instructions on
> >> the nuances of including the sed (or not), or skipping that command.
> >> Just want to make sure since this doesn't seem like the behavior you
> >> would want.
> >>
> >> My initial thought was some dependency missed, but it actually does find
> >> those files, it just doesn't find the expression so errors out non-zero.

That seems odd - I don't recall sed ever returning non-zero if a
change is not maade, only if the file doesn't exist.

A quick test on a local file, trying to change a word which is not
present, returns 0.

But now, a comment for anybody using just a console: "smart" quotes
in a sed may be evil!  My own console fonts, and some others, prefer
to NOT show a symbol for "this glyph is missing or invalid unicode"
(just before the closing parenthesis should be a highlighted
question-mark, U+FFFD, but it might just be a box, or a space �)
and the normal way to cope with the infernal smart quotes is to show
them as regular quotes.

It's a bit late now, but I would have preferred it if the
explanation mentioned smart quotes.

> 
> Sorry for the top post, I know better, but my email client didn't.  It
> does now. :)
> 
> Sorry if there is some user error.  I am copy and pasting using X.org
> copy/paste facilities (select to copy, middle-click to paste).  I have
> a binary chrome browser installed to make the build more pleasant with
> only twm as a window manager, and am building Gnome deps with xterm.
> 
> Copied text from rendered page shows up in terminal as:
>   sed -e 's//\"/' -e 's//\"/' \
> >     -i Source/WebCore/xml/XMLViewer.{css,js}
> From the raw page source, the XML element contains:
> sed -e 's/“/\"/' -e 's/”/\"/' \
> -i Source/WebCore/xml/XMLViewer.{css,js}
> 
> Perhaps “ and ” are too fancy for chrome to get into this
> type of copy/paste cleanly?
> 
In Xorg, I would expect the smart quotes to be available, but I
build with a decent set of TTF and OTF fonts, and not using xterm or
twm (nor luit) so I don't know for certain.

> I made a sample html file to test this theory:
> ---- BEGIN test.html ----
> <html>
> <head>
> <title>Test quotes</title>
> </head>
> <body>
> This is an ldquo;: “<br>
> This is an rdquo;: ”<br>
> This is a quot;: "<br>
> </body>
> ---- END test.html ----
> 
> After rendering in chrome, the output into the text editor run in xterm:
> This is an ldquo;:
> This is an rdquo;:
> This is a quot;: "
> 
> Sure enough, somewhere in the clipboard operation the first two
> "quotes" were ignored, while the third was handled.
> 
> I assume I would not have similar problems in lynx, but this is blfs
> so lots of folks are probably running X already by this point.
> 
> Is there a pressing need for the left and right quote flavors?  Is
> this a natural byproduct of the normal book rendering methodology, or
> some rendering framework?
> 
The left and right double quotes *are* the so-called smart quotes,
also known as '66' and '99' quotes, and they need to be changed to
regular ASCII '"'.

ĸen
-- 
`I shall take my mountains', said Lu-Tze. `The climate will be good
for them.'     -- Small Gods


More information about the blfs-support mailing list