Logseq is not a reliable note taking app

I am sure the team is working on fixing this but logseq is not just there yet. I spent almost a week into the rabbit hole of customising the logseq, learning the logseq only to realise some of my data is lost out of nowhere. Not to mention how slow logseq gets. UI/UX is pretty basic too in desktop as well as Phone app. Text on phone app looks big.

2 positive things about logseq: official team is active and very responsive and logseq has helpful and thriving community.

hope this helps any newcomer who wants to try logseq for the first time in upcoming months.

until you are sure people have stopped having data loss complains or if you have something not-so-important to try logseq with, i would not recommend logseq to anyone.

Welcome to Logseq and the community @prince! Sorry to hear your initial Logseq experience wasnā€™t good. I fully understand the frustration of data loss.

Logseq is still very much in beta, and some features are still in alpha. Thatā€™s no excuse, but Logseq isnā€™t labeled stable on purpose. Issues like data loss (common when using file-sync services like iCloud or Dropbox) are a big reason why Logseqā€™s architecture is being rewritten as we speak (see the announcement). Hopefully early next year Logseq will be a better choice for you.

And my two cents as a fellow PKM nerd: please donā€™t spend a week customizing a tool youā€™ve only just started using.

Having wasted hundreds of hours on tinkering with tools I never ended up using, I always recommend people first use a tool as-is for 1 to 3 months. Only once you know whether youā€™re going to stick with a tool is it worth spending hours tinkering and customizing it.

Just my opinion of course, but Iā€™ve been in your situation too many times and had the same issue when I just started out with Logseq. But the funny thing after three years of using this tool fulltime is that I barely use any plugins and have zero CSS tweaks in my graphs. Most of the customizations ended up being distractions anyway, not helping me be any more productive than just using vanilla Logseq.

In any case, thank you for your feedback and Iā€™ll make sure the dev team sees it as well.

13 Likes

What Iā€™m not saying is that your approach is somehow objectively incorrect, or that I disapprove of it, but from my experienceā€”and my first ā€œpersonalā€ computer was PDP-11/23, although I also used IBM/370 for larger compute loads, when it was available, so itā€™s likely that I have a lot of it simply by having hanged around computers for close to 45 yearsā€”learning a new thing by customising it is pregnant with the danger of destroying it, and only the thing alone by itself in the best case, or killing the learner at worst. This is especially true of kitchen nukers and acetylene welding torches: the former have voltages on the order of kilovolts that may fry anyone trying to learn it by disassembling and customizing instead of what theyā€™re built to fry, albeit in a different sense; the latter are prone to exploding horrifically, turning the learnerā€™s house into a freshly acquired lot of land with absolutely nothing on it when customised too much during learning.

Letā€™s step back to software, which is indeed safer but, just like everything else, no less prone to self-destruction from rough handling. Suppose a program has just 30 checkboxes as its option, and nothing else. Since there are 2Ā³ā°=1073741824 possible ways to configure this software. Thatā€™s a little bit over one billion, by just about 74 million. Itā€™s entirely unfeasible to expect that all this huge space of possibilities has been written a test for, as itā€™s unreasonable to maintain a billion tests. It is reasonable, however, to assume that some of the possible ways to customise the software have never been not only properly tested but even tried.

I spent a week of evening torturing Bing Copilot to tell me everything about software that would solve my very real problems, and, after reading reviews of the options then researching every one to a certain depth, eliminated most of them. I then read more in-depth articles and short videos on each of the remaining ones and settled on logseq as the most promising option. Then I went through a beginners course, then two deeper ones, played with the program along before erasing the journals that I created. Then I started working with it with a journal called Scratch that I also subsequently removed. I customised the date format to YYYY-MM-DD, changed no more than 5 shortcuts, and installed 1 (one) plugin, namely Bullet Threading, because the default theme with its light-gray text and spiderwebā€’thin block lines in nearly invisible, even to my slightly myopic eyes. When I noticed that I often manually create pages and copying too many references, I dug through the Marketplace and found two more plugins that solved the actual problem, Blockā€’Toā€’Page and enigmatically named logseq-swapblocks-plugin. This is where I am after first starting with logseq about 10 days ago. Everything is fast and smooth, although Iā€™m sure I could slow it down to a crawl by installing every plugin from the Marketplace. Iā€™m sure Iā€™ll find a way to make it more readable, but this can wait, as the primary function, namely sorting my notes, cannot wait any longer. I donā€™t mind squinting at the text for a week and then figuring out how to do that later, come next weekend.

Iā€™m just describing my experience; Iā€™m not suggesting in any way that my careful approach is the best, or is somehow better than that of yours. I just wanted to share how I approached the new software with an unfamiliar workflow and also very complex, so any tricks that I may unwittingly try may backfire in an entirely unexpected way.

Good luck with logseq! To me, it looks like itā€™s going to solve my problem by rescuing me from drowning in random paper and computer file notes. It certainly helps me already.

ā€”Cy

3 Likes

For me, the customization mostly did not put my data at risk. Once or twice because of a plugin that rebound keys. But the massive data loss that due to logseq sync. Whole pages were lost. Thankfully I think most of the data was restorable because I enabled Git. Git is hugely helpful for data recovery.

1 Like

In my case it has not been the loss of data, I pay sync and it works fine, the problem I have is that the tasks I have as ā€˜doingā€™ sometimes just disappear from the diary and stop showing, is it something that happens only to me or others also happens to them?

1 Like

If you have issues with Logseq Sync, please email support@logseq.com. I donā€™t read every post in the forum so sometimes I miss things, but I always keep a close eye on our support inbox.

Welcome to the community @alexchris! Itā€™s best to create a separate thread for this in Questions & Help. It also helps if you provide some more context. For example, where are the tasks exactly disappearing from? Is it from the page itself, or from the ā€œNOWā€ section at the end of the journal page? From your post I find it difficult to form a mental picture of whatā€™s exactly happeningā€¦

Well, luckyly I didn t have that experience yet. (2 years counting). I came to logseq because of the md-filebased architecture that guarantees a no lock-in environment. Sync I paid, tried and discarded due to the fact that I didn`t want all my stuff on the phone anyway. Running it local first and only, it is safe and secure from my perspective and allows to collect and ressource my notes and annotations to pdfs in a reasonable way.

I am not convinced that I will like the transition into a databased approach due to expected loss of the above mentioned liberty from lock-in.

What I miss though, is the option to install logseq effortlessly in an windows enterprise environment, that doesnt allow users to run code outside c:\program files while prohibiting them to write in said folder at the same time.

I can rollout the program package by script anyway. Last, closed shop environments require onPrem services to syncā€¦

I feel confident trusting logseq my journals and tasks as well as all the bits of kowledge that rain down on me day by day and even get them back if I need toā€¦

good work

Michael

Many programs install themselves per-user, under %LOCALAPPDATA% and, less commonly (probably incorrectly) under %APPDATA%. Among commonly used business programs, Zoom, Slack, Microsoft Teams, parts of Adobe infrastructure, Microsoft OneDrive all are installed this way. Of development and scientific tools, I know of VSCode, PerfView, LINQPad, Azure Functions framework, MSBuild Log Viewer, all JetBrain dev tools and executable packlets of Mathematica. Itā€™s a very common strategy for self-updating programs. The restriction doesnā€™t look amazingly reasonable. You may have a point to argue.

Unfortunately that is true for some programs. Still its wrong under enterprise security policies. Actually the Philosophy is, that userlevel installations do install into Appdata/local or Appdata remote. Quite some programs even ask you to choose if the installation is for ā€œall usersā€ on the given machine or just for the one trying to install it. Where in the first case, the installation commences to the Program Files folder and does need administrative rights to do so.

Which brings me back to the logic behind this security policy: There shall be no installation of executable code unless granted by security and administrative staff. Software is packaged and rolled out via Windows package management to reside in Program Files folder. Mostly even .msi files are provided with additional attributes enabled to configure the software on rollout.

On the other hand it is unsecure to install and use unvetted programs in an enterprise environment, which ist why security policy does not allow to run any executable that has not been installed the propper way to the expected location.

With logseq this is a luxury problem since I can, administrative rights provided, copy the respective folder from the nupkg to the required location and run it from there.

It would just be helpful, if the installation routine for Windows would offer the described option to install for all users and target the required program files folder.

tick me off on syncronisation, and Iā€™ll start a lecture about Cloud security policies, GDPR, confidentiality requirements and so on, not to forget the european - US gap and zero trust.

regards

Michael

To be fair, Logseq is not version 1 yet. It is still in beta stage. Things might break and data might be loss in beta stage. If you are looking for apps to store important/critical information, it is not recommended to use beta software.

Logseq is intentionally designed to minimalist with clean UI. Such design is a matter of preference, it may not be everyoneā€™s taste.

Hope you are able to recover the lost data.

2 Likes

How about you chip in some effort to make it reliable? If we get this service for free, I would not justify ranting at it.

1 Like

I had suffered similar issues, and somehow combined with other sync problems made the data unrecoverable. It is sad how open-source project struggle to be financial sustainable and keep dev going.
I tip Logseq every month, I love their approach to blocks and think it is an outstanding idea, I will still insist on using it because I lover their philosophy, however I share your worries and frustrations.
Is there other block-based outliner open-source having better performance and fewer bugs?
I think no.

2 Likes

FWIW - I have not experienced a single incident of data loss. Iā€™ve been a heavy user since early 2021 and hack on Logseq all the time, often writing some of my own custom JS code and have multiple times performed large bulk changes via scripts while LS is running. Iā€™ve never once used any sync service for any of my graphs which may be part of the reason why I have not experienced any sort of data loss.

2 Likes

I had data loss with Logseqā€™s own, paid sync service.
Moved to Obsidian, and I have NO data loss while using plain old free iCloud syncing.

So itā€™s weird to hear some of the excuses.

2 Likes

I donā€™t use logseq sync, Iā€™m scared of the data losses. Funny enough, I use gdrive for syncing data and didnā€™t lose any, but ai also make sure to use logseq only from one device at a time.

I do chip in monthly because I love the software. The database version scares me, but seems very interesting too at the same time.
Itā€™s also been a major blocker to new feature development, which Iā€™m looking forward to.

From my perspective, expanding simple query capabilities along with parameterized templates would make logseq incredible.
And the fix where querying an alias results in querying also the actual main tag

2 Likes