Fritz_Geiger

I have a (huge) lot of FF/NM's backups, starting from 3.6.28 times. And all the installed add-ons are unpacked. So I did search all files 'chrome.manifest' containing 'content '. Total 98 add-ons found. All entries are structured as content [add-on nickname] [value]. Typical values are: jar:chrome/[bla-bla-bla].jar!/[*] content/ content/[bla-bla-bla]/ chrome/content/ chrome/content/[bla-bla-bla]/ chrome/[bla-bla-bla]/content/ Rare but functioning values are: ./ chrome/ Unusual and non-functioning value, present only in Norwell, is: / The next question is - why slash works within archive and do not work in directory? Is it somehow connected with WXP rules that treats slashes? I think that it is within a step to off-topic. But just now the question is about why/how WXP behaves differently, including possible influence of some long-lasting FF/PM/NM unZIP-on-the-fly library. This version is not mentioned in http://forums.mozillazine.org/viewtopic.php?t=2860307 Is there any source that can be recomended to JustOff to include in CAA?

Great thanks! Last messages encourage me to try again. Usually I do all experiments with absolutely clean profile - my file NM_clean.CMD has start palemoon.exe -no-remote -profile .\pr_clean and the directory pr_clean is absent. But, in case of Norwell, as I am interested in processing my real/ripe file places.sqlite, I wanted to preserve it from default purge, so I run its install ("Open File..." bla-bla-bla.XPI) from the not-so-clean profile - there was my file user.js with abovementioned "places.history.expiration.interval_seconds" and "places.history.expiration.max_pages". And some other preferences, that I considered to be safe, f. e. "browser.cache.compression_level", "browser.shell.checkDefaultBrowser", "browser.startup.page" and so on, up to the famous "reversecowgirl_enabled". After an hour of tests I was ashamed. If "extensions.alwaysUnpack" is true (as I mentioned, sometimes I play with source code) when installing Norwell, it will not function. AFAIK, it is a first such error in FF life - at least in my FF life - add-on working from XPI only. Extremely strange.

Copy profile to RAMdisk at start (and save periodically to HDD) is a good method/advice, it was my default configuration in times of FF 3.6.28. Nowdays my places.sqlite are approx. 300 KB and 200 KB in two main/working browsers (planning to merge them, when issue with Norwell will be solved, or some SQlite knowledge will be obtained; so expecting near 0.5 GB - that is typical/average total RAM in my systems). So - "no" twice. First one for that my "deal with" is not equal to "keep fit" (includes, but not equal); my "deal with" is equal to "User configurable, by time and by volume". Second one for that I personally do not worship "fit"/skinny one, I prefer full-size/voluptuous, as Gianna Michaels. Well, I do not reject Stoya or Doris Ivy or Little Caprice... but I had to choose only one.

The reason is history chain. If nothing is deleted by URL (delete by time is irrelevant), it is possible to track all follow-ups/downs that took place. Whether it (storage/track) is desired by User, or not, depends on User. P. S. My friend Yuriy Dmitriev was sentenced to total 15 years in prison, and photos from his computer were used as evidences (of course, real reason is his human-rights-protection activity). I live in police state, so I fully agree with your concerns. But I do not agree with your solution. "Knowledge is power". I prefer to have an instrument to deal with history, rather than have no-usable/castrated history. And note, there was a sufficient instrument, at least for storage aspect - browser.history_expire*!

Well, it is normal in Google/Apple World, where everything is fixed to the floor by nails. In our world, it need to be selectable/configurable by User. [cit. from] https://blog.mozilla.org/tanvi/2013/04/10/mixed-content-blocking-enabled-in-firefox-23/#comment-547 "Hint: No one smart enough to use a browser besides what's bundled with their computer will want to use one that tries to hide controls from the users. Stop trying to act like Apple." To be sure, you can do periodic BAT/CMD snapshots. If your NM is started by CMD with something like - start palemoon.exe -no-remote -profile .\profile you can add 3 new strings below - sqlite3.exe places.sqlite .dump >> dump.sql [your preferred archivator creating timestamped-named archive containing dump.sql] del dump.sql The resulting file dump.sql is plain text, so it can be viewed by human and compared with previous/next one by any line-comparing program (text editors, f. e. MultiEdit, also have that function). Desired in "moz_places" and "moz_historyvisits" parts of dump.sql: nothing unique in older dump.sql (nothing deleted from history), new entries/lines in newer dump.sql.

Addition about history view. When FF does store some URL in history (and does not delete it, as depicted earlier), it stores every visit, with unique timestamp. But... Under menu item 'History' - 'Show All History' AKA upper part of 'Library' only the last visit is shown. And when that one-and-only-visible-to-User URL is deleted, all other, invisible-to-User, entries with that URL are deleted altogether. Putin! [Auto-Censoring Swear Words mode on] Well, there was Resistance - add-on Norwell History Tools by trumpai. I recommend to read it's description (on FF's inferiority) - available through Way-back Machine https://web.archive.org/web/201810/https://addons.mozilla.org/en-US/firefox/addon/norwell/ and Classic Add-ons Archive by JustOff caa:addon/norwell I tried this add-on in, AFAIR, FF 47, and it did show all. Direct link to the add-on https://ca-archive.us.to/storage/536/536280/norwell_history_tools-3.1.0.2-fx.xpi - but it does not function in NM 27 and NM 28 - installs, but no icon appears under menu item 'View' - 'Toolbars' - 'Customize' and Error console says 'Failed to load overlay' defined in file 'chrome.manifest' of the add-on. To my pity, I am not a programmer, and can not fix that add-on or create my own, my maximum in editing of the source code was achieved under guidance of the real author (add-on Clean Links by Diego Casorran, patched to remove/replace not only pairs 'parameter=value', but any sequence in the link).

You do not need an extension/add-on to increase the history duration. The starting entries in my NM 27 file user.js are: user_pref("places.history.expiration.max_pages", 2147483647); user_pref("places.history.expiration.interval_seconds", 2147483647); First one tells FF to keep 2^31-1 URLs/addresses in file places.sqlite (strictly speaking - sets max. number in/for its table 'moz_places'). Second one tells FF to check for expire once in 2^31-1 seconds (slightly more than 68 years). "God cares about those who are careful themselves" - said a nun rolling a second condom on a candle. -------------------------------- Background (_absolutely_ not required to read). --- Till, AFAIR, 2010 there were direct settings in FF: browser.history_expire_days browser.history_expire_days_min browser.history_expire_sites browser.history_expire_visits I used them at least in FF 3.6.28. Add-on 'Configuration Mania' under menu items 'Browser' - 'Other' has 'Max number of unique URLs to retain in history' with 'Manual' setting. But... all preferences browser.history_expire_* were removed; instead places.history.expiration.* appears/introduces. Horror story, including the birth of extension 'Expire history by days', in/at https://superuser.com/questions/1114637/does-firefox-ship-with-a-new-default-for-history-expires-after --- That new preferences places.history.expiration.* are unsufficient to control the activity of FF towards the file places.sqlite. A significant part of rules is hardcoded in the source code. They are evil. F. e., one rule _deletes_without_any_doubt_ everything that is more than 255 characters long. Other rule get rid of all redirects, thus breaking the history chain. And downloads, of course - who needs them in history? [sarcasm] Horror story, including the investigation of the source code, in/at https://superuser.com/questions/1269516/what-exactly-does-the-number-listed-under-places-history-expiration-transient-c --- My own search for 'places.history.expiration.' in the source code of PM 27. 'places.history.expiration.max_pages' 'places.history.expiration.transient_current_max_pages' 'places.history.expiration.interval_seconds' with comments and details present in palemoon27-master\toolkit\components\places\nsPlacesExpiration.js 38331 bytes 24.04.2021 22:38:44 'places.history.expiration.max_pages' 'places.history.expiration.interval_seconds' present in palemoon27-master\toolkit\components\places\tests\expiration\head_expiration.js 3356 bytes 24.04.2021 22:38:44 'places.history.expiration.max_pages' present in palemoon27-master\services\sync\tests\unit\test_history_tracker.js 6041 bytes 24.04.2021 22:38:44 palemoon27-master\toolkit\components\downloads\test\unit\test_history_expiration.js 3873 bytes 24.04.2021 22:38:44 palemoon27-master\toolkit\components\jsdownloads\test\unit\test_DownloadList.js 18338 bytes 24.04.2021 22:38:44 palemoon27-master\toolkit\components\places\tests\expiration\test_pref_maxpages.js 3718 bytes 24.04.2021 22:38:44 palemoon27-master\toolkit\components\places\tests\unit\test_telemetry.js 5432 bytes 24.04.2021 22:38:44 'places.history.expiration.transient_current_max_pages' present in palemoon27-master\toolkit\components\places\PlacesDBUtils.jsm 43791 bytes 24.04.2021 22:38:44 'places.history.expiration.interval_seconds' present in palemoon27-master\toolkit\components\places\tests\expiration\test_pref_interval.js 3923 bytes 24.04.2021 22:38:44 --- 'places.history.expiration.transient_current_max_pages' is a mirror, no need to set it directly. 'places.history.expiration.max_pages' 'places.history.expiration.interval_seconds' are absent in default about:config page, require to create them manually, or set in file user.js, that I recommend for all personal settings in order to be touched only by the User, distinguished from non-User, commented. --- I do not know about the situation in NM 28. I can not start to seriously use/tune it without knowing how to obtain a list of all preferences. In NM 27 under menu items 'Tools' - 'Web Developer' - 'Error Console' input in 'Code:' field top.opener.Services.prefs.getBranch("").getChildList("") for the list or top.opener.Services.prefs.getBranch("").getChildList("").length for the total number of preferences, press 'Evaluate', get more than 3 thousands. In NM 28 nothing happens. Can anybody tell what works in NM 28? Better not only names list, but with their settings. It means that FF does purge on its own decision, not in sync with User reasons/commands - FF decides, when and how. You are not the first person to express that thought.

Preference fast switch/toggle, without opening about.config. Historically I prefer more powerful/customizable solution than simple button. Since FF v3.6.28 I use add-on prefswitch (AMO/CAA number 6485). It puts (any number of any) preference switches on toolbar. When I tried it first time, I was somehow lost in the amount of possibilities (and not-very-regular interface) - downside of the fact that it can do everything. Alternative is PrefBar (AMO/CAA number 67148). P.S. Forum mulfunction in NM v27 again. In the same manner as it was approximately a year ago. No embedded references to post or picture are shown, blank space instead. And a thin, almost unnoticable reply editor line (vertically shrinked window). Tested in tuned 27.10.0 (32-bit) (2021-06-25) and in absolutely clean/fresh install with default profile without add-ons (unpacked and run with "palemoon.exe -no-remote -profile .\profile") 27.10.0.win32-git-20221112-5499b4356b-xpmod-ia32. To post this reply I used NM v28 on other computer, with SSE2.

I thought about changing UA before/during posting the problem, but decided to inform anyway, because even if I do solve it for me, it will still affect others and will not be fixed in the next NM version. Just now I set general.useragent.override.google.com same as default general.useragent.override.facebook.com - to Mozilla/5.0 (Windows NT 6.1; WOW64; rv:99.9) Gecko/20100101 Firefox/99.9 (Pale Moon) - and the problem is gone completely, Google search returns to normal layout. I prefer this unreal string. It directly/widely proclaimed - "UA is spoofed, because your site programmers/administrators are dumb or evil or both". Yes, I understand, that there are many cases, when not site authors, but site _engine_ authors (f. e., forum engine authors) are guilty, but most, if not all, NM built-in UA overrides are aimed on those who create that-f***ing-newest-Chrome-only-sites _themselves_. P. S. UA override _not_ helped for eBay item pictures. And there are some other (local-Russian) sites, that are affected with the same looking problem. F. e. at tv.mail.ru (TV channels charts) click on program name does nothing, instead of showing pop-up with detailed information.

Does this happens only for me? 1. Different NM 27.10.* versions (all ia32), including last 20220108, in WXP SP3. Unpack, run with "palemoon.exe -no-remote -profile .\profile" (i. e. clean/fresh install with default profile without add-ons). eBay item page. Any item. F. e., https://www.ebay.com/itm/133031616590 Item picture(s) left to the item name. Click on arrows (to scroll pictures left/right) or on image (to enlarge) does nothing. 2. Same NM versions + FirefoxPortableLegacy v52.9.0 20180621 Happened yesterday. Google search results page now looks like some mobile interface - thin center column, blocks/tiles, no page number selection.

While VLC v3 runs faster (smoother video) than v2 on both my WinXPSP3 systems, non-SSE and SSE2, v3 is unable to show streaming/http - "http error: invalid url" in log. Any v2 shows without problems this (or any other alive) URL. Sadly, I can not compare v3 log against v2 log - v2 does not produce log even directed by command line. So I stay at v2.2.1 (it reacts faster than later v2 versions when trying to stop the heavy/HD/4k video).

Thanks for your attention and clear answer. Now I plan to install the old-https-program (BTW, ReGet download manager) to other WinXPSP3 computer in my LAN, that has SSE2 and where HTTPSProxy can work. When ReGet on my current computer will not download some files (and the destination site uses https strictly, search&replace 'https' to 'http' in download list can not help), I will move the remaining queue to the computer with HTTPSProxy. It can be done effortless, and, in most cases, this is more time-effective than trying to manually edit requests and read negative server responses one-by-one. P. S. There are some proxy sites, that do conversion from "old" to "new" https, taking "old" requests from user and sending "new" to destination site. Like other free proxies, they do not last long. I used two Japan IPs In ReGet proxy settings, but they stopped approximately a month ago, and other free proxies provided by FireFox add-ons "One Click Proxy" and "Best Proxy Switcher", do not do the mentioned "conversion". So I started to search how implement "my own" proxy, "with blackjack and hookers" (c) Bender.

Is it possible to compile HTTPSProxy without SSE/SSE2? Actually I tested only the version included with launcher (v1.5.2 2018-11-06) - it does not run on non-SSE Duron, but does on SSE2 Pentium M.