contents.
When user switched from a page to another shorter the scrolling
position was not restored, pointing outside the actual text line
counts. This prevent the correct rendering of the page (a blank page
is shown instead) and crash the software if scrolling up.
Thanks op!
i got rid of slot support-text in message-window, so that class does
not maintain a state.
- changed behaivour of split-line
(split-line (format nil "~2%")) ; => ("" "")
That is an emtpy line is returned as empty string and not skipped
Thanks to people from IRC and cl-ppcre maintainer for help!
- [breaking changes] changed 'hooks:*before-prepare-for-rendering-message*
with 'hooks:*before-rendering-message-text*.
- there was a lot of problems related to type of an element of
'rows', they must be always a complex string
- moreover there was an incorrect positioning of the matching string
when character with column width >1 was present before the
string compatible with the regex.
line-based one.
The old method that packed text got from gemini stream in single event
was ugly and blocked the UI.
This version revert to a line by line rendering but it is optimized
and does not gets slow with the size of the text the window holds.
thread) the contents of a gemini file parsed so far without waiting
for the whole document to be read from the socket.
This feature should prevent blocking the client when the server
never close the socket (i.e. for streamed contents).