| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| | |
|
|\ \
| | |
| | |
| | | |
semantics close to [lrange] and [lreplace]. [dict merge] is also improved, but is not as strict.
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | | |
We do not want parsing an invalid dictionary to give errors about lists!
As compensation, we get greater precision in the errorcode.
|
| |\ \
| |/ /
|/| | |
|
|\ \ \
| | |/
| |/| |
|
| | | |
|
|\ \ \
| |/ /
| | |
| | | |
right on the next pass, don't forget the TCL_UTF_MAX padding demanded by Tcl_ExternalToUtf(). (Thanks for finding that, aku!) Fix the factorPtr management. It was just totaly wrong. The factor should be a ratio of the record of bytes read to the record of chars read. With those fixes, new test io-12.6 covers the "too many chars" code.
|
| |\ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
result in a panic.
The relevant function is ReadChars() (short RC in the following).
When the encoding and translation transforms deliver more characters
than were requested the iterative algorithm used by RC reduces the
value of "dstLimit" (= the number of bytes allowed to be copied into
the destination buffer) to force the next round to deliver less
characters, hopefully the number requested.
The existing code used the byte located just after the last wanted
character to determine the new limit. The resulting value could
_undershoot_ the best possible limit because Tcl_ExternalToUtf would
effectively reduce this limit further, by TCL_UTF_MAX+1, to have
enough space for a single multi-byte character in the buffer, and a
closing '\0' as well.
One effect of this were additional calls to ReadChars() to retrieve
the characters missed by a call with an undershot limit.
In the limit (sic) however this was also able to cause a full-blown
"Buffer Underflow" panic if the original request was for less than
TCL_UTF_MAX characters (*), and we are using a single-byte encoding
like iso-8859-1. Because then the undershot dstLimit would prevent the
next round from copying anything, and causing it to try and
consolidate the current buffer with the next buffer, thinking that it
had to merge a multi-byte character split across buffer boundaries.
(Ad *) For example because the previous call had undershot already and
left only such a small amount of characters behind!
The basic fix to the problem is to add TCL_UTF_MAX back to the limit,
like is done in all the (three) other places in RC setting a new
one. Note however that this naive fix may generate a new limit which
is the same as the old, or possibly larger. If that happens we act
very conservatively and reduce the limit by only one byte instead.
While I believe that this last conservative approach will never reduce
the limit to TCL_UTF_MAX or less before reaching a state where it
returnds the exact amount of requested characters I still added a
check against this situation anyway, causing a new panic if triggered.
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | | |
get it right on the next pass, don't forget the TCL_UTF_MAX padding
demanded by Tcl_ExternalToUtf(). (Thanks for finding that, aku!)
Fix the factorPtr management. It was just totaly wrong. The factor should
be a ratio of the record of bytes read to the record of chars read.
With those fixes, new test io-12.6 covers the "too many chars" code.
|
| | |\
| |_|/
|/| | |
|
| | | |
|
|\ \ \
| |/ /
| | |
| | | |
FlushChannel() which is its only caller. We need to discard the curOutPtr buffer as well, and not count on another pass through the loop to attempt to flush it (and raise the same failure again?).
|
| | |
| | |
| | |
| | |
| | | |
which is its only caller. We need to discard the curOutPtr buffer as well,
and not count on another pass through the loop to attempt to flush it
(and raise the same failure again?).
|
| |\ \ |
|
| | | | |
|
| | | | | |
| \ \ \ | |
|\ \ \ \ \
| | | |/ /
| | |/| | |
the close of a channel.
|
| | |\ \ \
| | |/ / /
| |/| | | |
|
| | | | |
| | | | |
| | | | |
| | | | | |
Several socket-14.* tests failing there, and those that pass are very slow
about it. Firewall or poor networking configuration may be playing a role.
|
|/ / / /
| | | |
| | | | |
a whole raft of test failures. WIP.
|
| |\ \ \
| | | | |
| | | | |
| | | | | |
connect monitoring. Bug [336441ed59]
|
| | | | | |
|
| | |\ \ \
| | |/ / /
| |/| | | |
|
| | | | | |
|
| | | | |\
| |_|_|_|/
|/| | | | |
|
| | | | |
| | | | |
| | | | |
| | | | | |
but don't resolve localhost to ::1 (and vice versa for IPv4 and 127.0.0.1).
|
| |_|_|/
|/| | |
| | | |
| | | | |
sufficiently strict.
|
|\ \ \ \
| | |_|/
| |/| |
| | | | |
PipeWatchProc(). When we are interested in both readable and writable events of a command pipeline channel, we only want the readable from the read end of the pipe, and the writable from the write end of the pipe.
|
| |/ /
| | |
| | |
| | |
| | | |
PipeWatchProc(). When we are interested in both readable and writable
events of a command pipeline channel, we only want the readable from
the read end of the pipe, and the writable from the write end of the pipe.
|
|\ \ \
| |/ /
| | |
| | | |
the flag BUFFER_READY.
|
| |\ \
| | | |
| | | | |
of the flag BUFFER_READY.
|
| | | | |
|
| | | |
| | | |
| | | | |
BUFFER_READY flag serves no necessary purpose, so it is removed.
|
| | | | |
|
| | | |
| | | |
| | | |
| | | | |
the OS networking machinery gets backed up and blocks.
Added several TODO comments on potential simplifications.
|
| |/ / |
|
| | | |
|
|\ \ \
| |/ /
| | |
| | |
| | | |
platforms into the trunk. For details see the merged revision and its
ancestor.
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
regarding the handling of wrapped dynamic libraries.
The basic flow of operation is to copy such libraries into a temp
file, hand them to the OS loader for processing, and then to delete
them immediately, to prevent them from being accessible to other
executables. On platforms where that is not possible the library is
left in place and things are arranged to delete it on regular process
exit.
An example of the latter are older revisions of HPUX which report that
the file is busy when trying to delete it. Younger revisions of HPUX
have changed to allow the deletion, but are also buggy, the OS loader
mangles its data structures so that a second library loaded in this
manner fails.
More recently it was found that Linux which is usually ok with
deleting the file and gets everything right shows the same trouble as
modern HPUX when the "docker" containerization system is involved, or
more specifically the AUFS in use there. Deleting the loaded library
file mangles data structures and breaks loading of the following
libraries. For a demonstration which does not involve Tcl at all see
the ticket
https://github.com/dotcloud/docker/issues/1911
in the docker tracker.
This of course breaks the use of wrapped executables within docker
containers.
This commit introduces the function TclSkipUnlink() which centralizes
the handling of such exceptions to unlinking the library after unload,
and provides code handling the known cases. IOW HPUX is generally
forced to not unlink, and ditto when we detect that the copied library
file resides within an AUFS.
The latter must however be explicitly activated by setting the define
-DTCL_TEMPLOAD_NO_UNLINK during build. We still need proper configure
tests to set it on the relevant platforms (i.e. Linux).
The AUFS detection and handling can be overridden by the environment
variable TCL_TEMPLOAD_NO_UNLINK which can force the behaviour either
way (skip or not). In case the user knows best, or wishes to test if
the problem with AUFS has been fixed.
|
| | | |
|
| | |
| | |
| | |
| | | |
[dict create] to explicitly state that it returns the new dictionary.
|
| | | |
|
| | |
| | |
| | |
| | | |
value.
|
| | | |
|
| | | |
|
| | | | |
| \ \ | |
|\ \ \ \
| | |/ /
| | | |
| | | | |
particular not allowing them to leak between multiple layers of a stacked channel. Much common code refactored into ChanRead().
|
| | |\ \
| | | | |
| | | | |
| | | | | |
particular not allowing them to leak between multiple layers of a stacked
channel. Much common code refactored into ChanRead().
|