1 Some Privoxy-related tasks, sorted by the time they
2 have been added, not by priority.
4 The latest version should be available at:
5 https://www.privoxy.org/gitweb/?p=privoxy.git;a=blob_plain;f=TODO;hb=HEAD
7 There's work in progress to fund development on these items using
8 donations. If you want to donate, please have a look at:
9 https://www.privoxy.org/donate
11 1) Add more regression tests. Filters should be tested automatically
12 (variables too). Could probably reuse large parts of Privoxy-Filter-Test.
13 Note that there is currently work in progress to leverage curl's
14 test suite, patches have been submitted upstream:
15 https://curl.se/mail/lib-2014-06/0070.html
16 https://curl.se/mail/lib-2021-01/0068.html
18 3) Fix some more XXX: comments.
20 6) Remove actions that aren't needed anymore:
22 content-type-overwrite should probably stay as it's also
23 used by some of the CGI pages (XXX: name them).
25 crunch-client-header and crunch-server-header should probably
26 go, their only advantage is that their search strings can be
27 controlled through the CGI pages, other than that they only
30 crunch-if-none-match can be replaced with a header filter.
32 prevent-compression has a misleading name and could
33 be replaced with a header filter.
35 7) force-text-mode has a stupid name and should probably
36 be renamed to force-filter-mode.
38 8) handle-as-empty-document and handle-as-image should
39 be merged to something like handle-as{something} to
40 prevent them from being activated at the same time.
42 10) There's a bug in the CGI editor that turns the
43 first section's "Insert new section below" into
44 a "Insert new section above" button.
46 11) CGI templates should use semantically-correct HTML
51 12) Support pipelining for outgoing connections.
53 19) enable-forward-fallback. Syntax? Suggested by K.R.
55 21) User Manual delivery doesn't accept multiple slashes. Should it?
57 22) Verify action files properly (Including arguments) and
58 act accordingly (should probably intercept all requests
59 with a "Invalid option foo detected" CGI page).
61 23) Do the same in case of syntax errors in the configuration file,
62 instead of just exiting or ignoring the problem.
64 25) Handle multiple filters with the same name better. Reject them?
66 26) Let show-url-info detect clearly invalid URLs.
68 27) Make errno logging less thread-unsafe.
69 Verify that it's really an improvement.
71 28) Don't take default ports in case of invalid forwarding ports.
73 31) If a string action foo is disabled csp->action->string[ACTION_STRING_FOO]
74 doesn't necessarily contain NULL, but may contain the string of an
75 enabled foo action in an overruled section. Is it a bug? Does it matter?
77 32) In case of forwarding failures with socks port == 9050,
78 show extra info about Tor (the whole FAQ entry?).
80 36) Unload unused action files directly, even if they are
81 disabled without replacement.
83 38) In the final results, explicitly list disabled multi actions
84 with their parameters. Not as trivial as it sounds.
86 40) When running in daemon mode, Privoxy's working directory is '/'
87 which means it may not have permissions to dump core when necessary.
88 Figure out a way to solve this. Introduce a cwd config option?
90 41) Change documentation framework to one that works cross-platform.
91 Evaluate WML and txt2tags.
93 42) Add a DTrace USDT provider. Now that FreeBSD has userland DTrace
94 support there's no longer any reason not to.
96 43) Write a tool to check URL patterns against URLs in the log file.
97 This could be included in Privoxy-Regression-Test.
99 44) Privoxy-Log-Parser: Consider highlighting "Connection" in:
100 23:13:03.506 283b6100 Header: Replaced: 'Connection: Keep-Alive' with 'Connection: close'
102 50) Investigate possible PCRS template speedup when searching
103 macros with strstr() before compiling pcrs commands.
104 Investigated, needs some restructuring but is probably worth it.
106 51) Make user-manual directive more generic to allow serving the FAQ
107 and files from user-specified directories. Consider changing the
108 port for "same origin policy" issues.
110 53) Find a more reliable hoster. Involves finding out what our
111 requirements are and which SF alternatives fulfil them.
112 It would probably also make sense to look into what other
113 projects did when migrating away from SF.
115 2014-05: Work in progress. Hosting wish list at the end
116 of this file. Looks like most of the other projects
117 that left SF had lower standards and moved to hosters
118 that don't come close to sattisfying the requirements.
119 2016-03: The website has been moved away from SF infrastructure
120 and is also available through https:// now.
121 2016-04: Server rent for a year has been sponsored by ChameleonJohn.
122 2016-04: The SF mailing lists have been deprecated, the new ones
123 are available at: https://lists.privoxy.org/
125 Interested donors: 1.
127 58) Move more template strings from the code into the actual templates.
129 59) Import the German template translation.
131 60) Ask the Russian translators for input on how to make their
134 61) Consider (optionally?) skipping the hostname comparison when
135 checking if a connections that goes to a HTTP proxy can be reused.
136 Do all HTTP proxy support that? Is it worth it?
138 63) Reject clearly too large requests earlier?
140 64) Use proper copyright attribution. "Privoxy Developers"
143 65) Polish Website. Probably involves ditching the Docbook
144 mess. There are already several threads in the mailinglist
145 archives about this. See also #41.
147 66) Stop hard-coding the number of action and filter files.
149 67) Clean up source code directory layout.
151 68) Use standard make syntax so we don't depend on GNU make.
153 69) Update autoconf setup (or move away from it).
154 Unfortunately the autoconf files can't be simply updated
155 due to license issues:
156 https://lists.privoxy.org/pipermail/privoxy-devel/2016-April/000008.html
158 70) If the server connection is reset but the headers are
159 received, consider passing the mess to the client instead
160 of showing the connect-failed template. Relates to #2698674.
162 74) Let Privoxy-Regression-Test optionally check that action
163 sections which disable actions actually are preceded by
164 sections that enable said actions.
166 75) Create a tool that creates Privoxy action (and filter?) files
167 out of adblock files. Could be implemented as option for
168 url-pattern-translator.pl. Before doing that, the already
169 existing solutions should probably be evaluated to see if
170 they do the job already or could be improved.
172 76) Cache DNS responses. Note that this has been requested
173 several times by users, but is not a developer priority.
174 If you care about this, feel free to submit patches.
176 77) Allow to configure the IP address used in outgoing connections.
178 78) Allow to optionally use pcre's DFA algorithm.
180 79) Evaluate pcre alternatives.
182 82) Detect if the system time goes back in time let the user
183 know if it caused any connections to get closed.
185 85) Once #51 is done, write a script that populates a directory with
186 various common third-party icons (stumbleupon.png, facebook.png ...)
187 and redirect requests for them to Privoxy.
189 86) Add a server-body-tagger action. This is trivial as as all the
190 functionality required to do it already exists.
192 87) Add a client-body-tagger action. Work in progress.
194 88) Investigate if there's a Perl module that Privoxy-Regression-Test
195 could optionally use to keep connections alive, preferably while
196 requiring less forks at the same time.
198 89) When multiple block actions apply, consider showing all the block
199 reasons on the blocked page that haven't been overruled, not just
202 91) Add an optional limit for internal redirects. It would probably
203 be reasonable to default to a limit of one and showing an error
204 message if the request for the redirect URL would be redirected
207 92) The statistics currently aren't calculated correctly by Privoxy
208 as each thread is only counted as one request which is no longer
209 correct. This should be fixed, or the statistic code removed.
210 Privoxy-Log-Parser's provides more detailed statistics, anyway.
212 93) Add a config directive to let Privoxy explicitly request either
213 IPv4 (or IPv6) addresses, even if the system supports both.
214 Could be useful as a workaround for misconfigured setups where the
215 libc returns IPv6 addresses even if there's no IPv6 connectivity.
217 94) Add a config directive to let Privoxy prefer either IPv4 (or IPv6)
218 addresses, instead of trusting the libc to return them in an order
219 that makes sense. Like #93, this could be useful as a workaround
220 for misconfigured setups.
222 96) Filters should be easier to look up. Currently get_filter() has to
223 go through all filters and skip the filter types the caller isn't
226 98) When showing action sections on the CGI pages, properly escape
227 line breaks so they can be copy&pasted into action files without
230 99) Figure out a mechanism through which a user can easily enable
231 site-specific action sections that are too aggressive to be
232 enabled by default. This could be similar to the presettings
233 in default.action, but could also be just another action file
234 that isn't used by default.
236 100) Create a cross-platform Privoxy control program and retire
237 the win32 GUI. Integrate support for Privoxy-Regression-Test,
238 Privoxy-Log-Parser, Privoxy-Filter-Test, uagen and similar tools.
239 Interested donors: 1.
241 102) Add an include directive to split the config file into several parts.
243 103) Potential performance improvement for large action files:
244 when figuring out which actions apply, check the action bit mask
245 before pattern matching and skip section that wouldn't modify the
246 actions already set. To increase the impact the sections would have
247 to be applied in reverse.
249 104) The code to modify global_toggle_state should be factored out into
250 a separate function. Currently we mess with it in three different
251 files, but only in w32log.c the tray icon is explicitly set.
252 The logging is inconsistent as well. For details see #3525694.
254 106) actionlist.h should be embedded in a way that causes less text
257 107) Support more pcrs variables, for example $destination-ip-address
258 and $source-ip-address.
260 108) Allow to use a somewhat random string instead of PRIVOXY-FORCE.
262 109) Let log_error() support the format specifier %S which should
263 work like %s but escape new lines like %N. This would be useful
264 to log the result of header filters which may inject new lines.
266 110) Add a global-buffer-limit directive that roughly limits how
267 much malloc'ed memory Privoxy will use and can potentially
268 be smaller than (buffer-limit * max-client-connections).
270 111) Reject requests if hosts and ports in request line and Host
271 header don't match (before filters have been applied).
273 112) If a header filter is used to inject another header by inserting
274 a \r\n (undocumented feature), detect it and split the headers so
275 following header actions do not treat them as a single string.
276 Alternatively add another header injection mechanism.
278 113) Log statistics upon receiving a certain signal (SIGINFO or SIGUSR1).
280 114) Properly deal with status code 100. The current "Continue hack"
281 can cause problems for gpg when uploading keys through Privoxy.
283 115) Add ICAP (RFC 3507) support. FR #3615158.
285 116) Due to the use of sscanf(), Privoxy currently will fail to properly
286 parse chunks whose size can't be represented with 32 bit. This is
287 unlikely to cause problems in the real world, but should eventually
288 be fixed anyway. See also:
289 https://bugzilla.mozilla.org/show_bug.cgi?id=959100
291 118) There should be "escaped" dynamic variables that are guaranteed
292 not to break filters.
294 120) Add an option to limit pcre's recursion limit below the default.
295 On some platforms the recursion limit doesn't prevent pcre from
296 running out of stack space, causing the kernel to kill Privoxy
299 121) Add HTTP/2 support. As a first step, incoming HTTP/1.x requests
300 should be translated to outgoing HTTP/2 requests where possible
301 (and if desired by the user).
302 Interested donors: 1.
304 122) Allow customized log messages.
306 124) Add support for the "lightweight OS capability and sandbox framework"
307 Capsicum. http://www.cl.cam.ac.uk/research/security/capsicum/
308 Interested donors: 1.
310 125) Allow clients to HTTPS-encrypt the proxy connection.
311 Interested donors: 1.
313 126) Run the Co-Advisor HTTP compliance tests, evaluate the results,
314 fix the compliance issues that aren't by design and document
316 Note that Privoxy developers qualify for free account upgrades:
317 http://coad.measurement-factory.com/details.html#pricing
319 127) Add "real" CGI support (serve program output instead of forwarding
320 the request). The work is mostly done due to +external-filter{}.
322 128) Add a config directive to control the stack limit.
324 129) Completely implement RFC 7230 4.1 (Chunked Transfer Coding).
325 Currently Privoxy doesn't properly deal with trailers which
326 are rarely used in the real world but should be supported anyway.
328 130) Move header_tagger() out of the parser structs and let it execute
329 taggers one-by-one against all headers so the header order has less
330 influence on the tagging result. As a bonus, dynamic taggers would
331 have to be compiled less often.
333 131) The handle-as-empty-doc-returns-ok directive should be replaced with
334 an action so the behaviour can be enabled on a per-request basis.
335 Interested donors: 1.
337 133) Consider allowing bitcoin donations.
338 Interested donors: 2.
340 134) Track the total number of bytes written to and received from a socket.
342 135) Add OpenBSM audit support.
344 136) Make builds reproducible.
346 138) Bring back the scripts to provide actions file feedback.
348 Once upon a time (~2003) there were scripts on the webserver
349 to make reporting action file feedback more convenient for the
350 user and the actual reports more useful for the developers.
351 They have been unusable for years and have thus been disabled,
352 but making the reporting mechanism available again would be a
355 140) Toggling Privoxy off currently also disables stuff that
356 probably shouldn't be affected (such as actions like
357 forward-override). Investigate and fix or document.
359 143) Add support for OpenBSD's pledge feature once it's stablelized.
360 This should be a lot less work then #124.
362 146) Allow to save the internal client tag state to disk and
363 load it after restarts.
365 147) Improve "Building from Source" section in the user manual.
366 A common problem seems to be that it's not obvious to non-technical
367 users how the listed dependencies can be installed on the commonly
368 used platforms. Adding a couple of examples should also be useful for
369 technical users (like Privoxy developers) who want to install or test
370 Privoxy on platforms they are not familiar with.
372 148) Add a config directive to change the CGI_SITE_2_HOST
373 (default: config.privoxy.org).
375 If Privoxy is used as reverse proxy or intercepting proxy without
376 getting intercepted requests, error pages created from default templates
377 currently can result in client requests to config.privoxy.org on the
378 Internet which may not be desirable.
380 150) Add blocklistd support (https://man.netbsd.org/blocklistd.8).
382 151) Let the dok-tidy target work cross-platform without introducing
383 a ton of white-space changes that hide the content changes.
385 152) Fix CSS references in the website documentation.
386 For many pages p_doc.css is specified twice using different paths.
387 Usually at least one works, but not all of them do and the
388 duplicated requests are pointless even if they don't end up with
391 153) Catch SIGINT and use it to close the listen socket, serve
392 remaining connections and shut down. This would allow higher
393 uptime and make testing more convenient.
395 154) Underline links in docs and cgi pages. More precisely,
396 don't mess with the browser defaults for link underlining.
398 155) The sig_handler() shouldn't call log_error().
399 While it isn't known to cause actual problems in normal operation,
400 it's technically incorrect and causes crashes when running in
403 156) Reject socks requests with an explicit error message similar
404 to the one used for ftp. Motivation:
405 https://lists.privoxy.org/pipermail/privoxy-users/2017-March/000195.html
407 158) Use a single thread to wait for new requests on reused client connections.
408 Currently the thread that handles the first request on a connection
409 stays responsible for the client connect until it gets closed.
410 In case of lots of idle connections lots of waiting threads are used.
411 While it's conceivable that this ineffiency is irrelevant from a
412 performance point of view, using a single thread should reduce Privoxy's
413 memory footprint a bit which may be noticeable in case of multi-user setups
414 with hundreds of idle connections.
416 161) Properly support requests with chunked transfer-encoding with https inspection.
418 162) When https inspecting, delete generated keys and certificates if
419 the connection to the destination could not be established.
420 Makes silly DoS attacks slightly more complicated.
422 163) Use subdirectories in the certificate-directory to lower the number
423 of files per directory.
425 164) Evaluate switching from pcreposix(3) to pcre's native api
426 for URL matching which allows to compile the patterns once
429 165) Add a max-connections-per-client directive.
431 167) Set up a public Privoxy-Filter-Test instance.
433 168) Add a privacy policy.
435 169) Preserve all relevant copyright and license statements in binary
436 packages we distribute.
438 170) Serve the ca-cert-file through the CGI interface so clients
439 can conveniently import it (insecurely).
441 171) Create a "view page using Privoxy" website where users can input
442 an URL and get a screenshot of a browser fetching the URL
445 172) Create a public git repository for Privoxy-Filter-Test.
447 173) Document Privoxy's governance model.
449 174) Let the Tor Onion Service for the privoxy website
450 serve gitweb and the git repository as well.
452 175) Add more screenshots to the documentation and website.
454 176) Find a new fiduciary sponsor as a replacement for Zwiebelfreunde e.V.,
455 so that we can continue to receive tax-deductible donations in Europe.
457 177) Support https-inspection for intercepted requests.
459 178) Warn on http://config.privoxy.org/client-tags if a Tag name
460 hasn't at least one matching action section.
462 179) Add a add-server-header{} action to add headers to the response
463 sent to the client (including responses generated by Privoxy itself).
465 180) Add support for GnuTLS.
467 181) Allow to upgrade an http request to https behind the
468 client's back using a client-header filter.
470 182) Before enforcing the client-header-order, check that the
471 client headers actually need sorting. Should reduce log
472 messages and memory allocations.
474 183) Properly deal with proxy responses that arrive in multiple pieces
475 when https inspecting while using a forwarding proxy.
477 184) Add support for wolfSSL. Work in progress, expected to be
478 committed after the 3.0.34 release. Funded with donations
479 made to the Privoxy project.
481 185) The mbedTLS and OpenSSL versions of generate_host_certificate()
482 should only be called when necessary and the check should be
483 done without holding the certificate mutex.
485 186) Privoxy should handle "OPTIONS *" requests properly.
487 187) There should be a convenient way to see the versions of
488 the libraries Privoxy is using.
490 188) In the windows config.txt file, add the line
491 user-manual ./doc/user-manual/
496 189) Bring back binary packages for macOS, preferably for both Intel and M1.
497 The first step would be getting at least one build system, either
498 donated or bought with donations earmarked for this.
499 Interested donors: 0.
501 190) The socks5 authentication code should send user name an password
502 seperately or we should increase the cbuf size to allow longer
503 user names and passwords.
505 191) The cipher-list directive should be split into cipher-list-server
506 and cipher-list-client.
508 192) The client TLS contexts should probably be shared among threads
509 to spend less time and memory loading the root certificates.
511 193) Use SHA256 instead of MD5 for the host hash used when generating file
512 names for host certificates and keys.
514 194) There should be a way to force gif deanimation if the server does not
515 declare the content as gif.
517 195) We should probably cache the server TLS contexts.
519 196) Investigate if it's worth adding an optional mutex for the CGI handler.
520 Could reduce memory use and increase performance on single core systems
523 197) Investigate if parts of Privoxy should get optional replacements
526 198) Add a config directive that prevent's IP addresses from being logged
527 (when logging is enabled).
529 199) In actions.c the "#define DEFINE_ACTION_ALIAS 0" lines should probably
530 be changed to "#undef DEFINE_ACTION_ALIAS" or removed.
532 200) Add a config directive that causes Privoxy to remove all
533 host certificates before exiting.
535 201) Add an action to change the trusted-cas-file for a section.
536 This should be useful in countries where a person-in-the-middle
537 attack is known to happen on some domains but should not be tolerated
538 on others. It would also allow to limit the accepted CA certificates
539 for given domains instead of accepting all that are specified with
540 the trusted-cas-file directive.
542 202) Allow to use multiple log files with different debug settings.
543 Suggested by Jonathan Marten in FR#607.
545 ##########################################################################
547 Hosting wish list (relevant for #53)
552 - Mailinglists (Mailman with public archives preferred)
553 - Webspace (on a Unix-like OS that works with the webserver targets
556 - Commit mails (preferably with unified diffs)
558 (Unsorted) details to look at when evaluating hosters:
560 1. Preferably no third-party ads and trackers.
561 External images, CSS and JavaScript may count as trackers
562 but texts like "supported by company XYZ" may be acceptable.
564 2. JavaScript should be optional or not used at all.
566 3. Services we don't need shouldn't be enabled anyway.
567 (We currently don't use Web forums, wikis, surveys etc.)
569 4. It would be preferable if the hoster didn't have a bad track
570 record as far as user experience, security and privacy are
571 concerned and if the terms of service are "reasonable" and
572 haven't changed too often in the past. Updates in the past
573 should have been improvements and not regressions.
575 5. It would be preferable if most of the server administration
576 is done by a trusted third-party (or at least not a lot of work
579 6. The server(s) should be located in a country with laws we can
580 understand and follow (or at least not unintentionally violate).
582 7. A server location in a country with some kind of due process
583 and strong data protection laws (at least on paper) would be
586 8. Given that Privoxy is a free software project it would be
587 preferable if the hoster would use free software where possible.
589 9. Migrating away from the hoster in the future without losing
590 any important data should be possible without writing web