Hi,

there's one annoying bug I'm experiencing with geany. These are the steps to reproduce it:

1. Open a file and scroll down a bit. 2. Open the file again e.g. by double clicking the filename in the file browser plugin. 3. Click somewhere in the editor.

What happens is that the editor moves down by several lines and selects the text in between what was the position of the cursor before and the position of the cursor after the movement.

I have bisected the commit that causes the problem with git and it was introduced by the following commit:

SVN: 4151 git: f2771531403ef0094274f8c5654f1742d7e88918

2009-09-04 Apply patch from Thomas Martitz to improve sidebar type-ahead code: Use gtk_notebook_set_current_page() instead of document_open_file_full() when choosing an item from the documents list. Avoid using goto in document_open_file_full().

(+ friend commits 4158 that fixes geany crash and 4197 that prevents the cursor from moving to the beginning of the file when reloading)

Any idea what might be the problem? Sorry for not providing a patch that fixes it but I thought that someone might know better what's wrong (I actually don't even know the purpose of the patch).

There's one more problem related to wrong cursor position. After updating to Scintilla 2.12 (also bisected) when a file is opened from a message window at certain line (e.g. by clicking at ./foo.c:80: resulting from find in files) the document position is different when the file is loaded for the first time (the first click in the message window) from the position when the document is reloaded for the second time (the second click in the message window). This is possibly just a Scintilla 2.12 bug but might be also related to the above bug (why does the editor behave differently when the document is opened for the first time from the case when it's (re)opened for the second time?).

Also it happened to me several times that when I opened a file by clicking a result of FIF in the message window, the found line was below the current viewport. I haven't encountered this recently so maybe it has been fixed in the meantime, just reporting it for completeness.

Cheers,

Jiri